0% found this document useful (0 votes)

214 views40 pages

Smart Grid Future PDF

Uploaded by

Pooja

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

214 views40 pages

Smart Grid Future PDF

Uploaded by

Pooja

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 40

This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2019.2894819, IEEE Access

Date of publication xxxx 00, 0000, date of current version xxxx 00, 0000.
Digital Object Identifier 10.1109/ACCESS.2017.Doi Number

Application of Big Data and Machine Learning

in Smart Grid, and Associated Security
Concerns: A Review
Eklas Hossain1, Senior Member, IEEE, Imtiaj Khan2, Fuad Un-Noor3, Sarder Shazali
Sikander4, and Md. Samiul Haque Sunny3
1
Department of Electrical Engineering & Renewable Energy, Oregon Tech, Klamath Falls, OR 97601, USA
2
Department of Electrical and Electronic Engineering, Bangladesh University of Engineering and Technology, Dhaka, Bangladesh
3
Department of Electrical and Electronic Engineering, Khulna University of Engineering and Technology, Khulna 9203, Bangladesh
4
National Université of Science and Technologie Islamabad, Pakistan

Corresponding author: Eklas Hossain (e-mail: eklas.hossain@oit.edu).

ABSTRACT This paper conducts a comprehensive study on the application of big data and machine
learning in the electrical power grid, introduced through the emergence of the next-generation power system
– the smart grid (SG). Connectivity lies at the core of this new grid infrastructure, which is provided by the
internet of things (IoT). This connectivity, and constant communication required in this system, also
introduced a massive data volume that demands techniques far superior than conventional methods for
proper analysis and decision-making. IoT integrated SG system can provide efficient load forecasting and
data acquisition technique, along with cost effectiveness. Big data analysis and machine learning techniques
are essential to reap these benefits. In the complex connected system of SG, cyber security becomes a
critical issue; IoT devices and their data turning into major targets of attacks. Such security concerns and
their solutions are also included in this study. Key information obtained through literature review are
tabulated in corresponding sections to provide a clear synopsis; and the findings of this rigorous review are
listed to give a concise picture of this area of study and promising future fields of academic and industrial
research, with current limitations with viable solutions along with their effectiveness.

INDEX TERMS Big Data Analysis, Cyber Security. IoT, Machine Learning, Smart Grid.

VOLUME XX, 2017 1

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2019.2894819, IEEE Access

LIST OF ABBREVIATIONS CCM Cipher block chaining-message authentication code

IoT Internet of things GCM Galois/counter mode
ML Machine learning GMAC Galois message authentication code
SG Smart grid HMAC Hash-based message authentication code
DER Distributed energy resources CFB Cipher feedback
DEM Dynamic energy management CBC Cipher-block chaining
CPL Constant power load ECB Electronic codebook
MOSFET Metal-oxide-semiconductor field-effect transistor XTS XEX-based tweaked-codebook mode with ciphertext
HDFS Hadoop file system stealing
LPRF Low power radio frequency TDES Triple data encryption standard
OFDM Orthogonal frequency-division multiplexing TECB TDEA electronic codebook
HAN Home area network TCBC TDEA cipher-block chaining
NAN Neighbor area network TCFB TDEA cipher feedback
WAN Wide area network TOFB TDEA output feedback
HG Home gateway CTR Counter-mode
ESP Energy service provider IaaS Infrastructure-as-a-service
PDC Phasor data concentrator SaaS Software-as-a-service
PMU Phasor measurement unit PaaS Platform-as-a-service
TCP/IP Transmission control protocol/Internet protocol DaaS Data-as-a-service
WAMS Wide area measurement system CVM Core vector machine
UDP User datagram protocol  PMU Micro phasor measurement unit
NRECA National rural electric cooperative association PPMV Power plant model validation tool
NLP Natural language processing FRAT Flight risk assessment tool
PCA Principal component analysis SVM Support vector machine
K-NN k-nearest neighbors PQ Power quality
ANN Artificial neural network SEMMA Sample, Explore, Modify, Model, Assess
CFD Computational fluid dynamics CRISP- Cross Industry Standard Process for Data Mining
CxO Corporate officer DM
BDC Billing and debt collection ELM Extreme learning machine
SCADA Supervisory control and data acquisition ANFIS Adaptive neuro-fuzzy inference system
PLC Programmable logic controller RVM Relevance vector machine
EMS Energy management system EMD Empirical mode decomposition
DMS Distribution management system GMR Generalized mapping regressor
CPS Cyber physical system CRO Coral reef optimization
HEMS Home energy management system iSSO Improved simplified swarm optimization
WLAN Wireless LAN HAP Hybrid swarm technique
MPO Meter point operator PSO Particle swarm optimization
DoS Denial of service ACO Ant colony optimization
DDoS Distributed denial of service TPSD Three-phase signal decomposition
FDIA False data injection attack WRELM Weighted regularized extreme learning machine
MITM Man-in-the-middle attack SSA Seasonal separation algorithm
RF Radio frequency FEEMD Fast ensemble empirical mode decomposition
NERC North American Reliability Corporation VMD Variation mode decomposition
API Application program interface PACF Partial autocorrelation function
MSE Mean squared error MLP Multilayer perceptron
HMM Hidden Markov model LMS Least median square
PID Proportional integral derivative CRO-ELM Coral reef optimization – extreme learning machine
RMT Random matrix theory GPR Gaussian process regression
DNN Deep neural network LVQ Learning vector quantization
SNN Shallow neural network SOM Self-organizing map
DSHW Double seasonal Holt-Winters SVR Support vector regression
GPU Graphics processing unit
UCSD University of California San Diego
PV Photovoltaic
MV/LV Medium voltage/Low voltage I. INTRODUCTION
MPPT Maximum power point tracking The electrical power system is poised to move towards the
AMI Advanced metering infrastructure next-generation smart grid (SG) system, and thus this topic
ISMS Information security management system
SoGP Standard of good practice
has acclaimed significant attention in the research
IACS Industrial automation and control systems community [1-7]. SG is the integration of information and
AES Advanced encryption standard digital communication technologies with power grid systems
TDEA Triple-data encryption algorithm to enable bi-directional communication and power flow that
DSS Digital signature standard
DSA Digital signature algorithm
can enhance security, reliability, and efficiency of the power
RSA Rivest, Shamir, and Adleman system [8-10]. Smart grid solutions aim at calculation of
ECDSA Elliptic curve digital signature algorithm optimum generation-transmission-distribution pattern and
SHA Secure hash algorithm storing power system data. For the growing concern about
CMAC Cipher-based message authentication code

Author Name: Preparation of Papers for IEEE Access (February 2017)

environment along with efficient generation and distribution,

distributed energy resources (DER) with smart microgrid can
be a potential solution [11]. It can be said that distributed
smart microgrid can bring additional benefits for global
power system planning [12]. In other words, SG is the
integration of technologies, systems and processes to make
power grid intelligent and automated [13]. Fig. 1 shows basic
constructions of conventional grid and smart grid to
demonstrate their dissimilarities. Unlike the unidirectional
power flow in the conventional system, power and
information flow between the generation and distribution
sides in a bidirectional manner.

(a)

(b)
FIGURE 1. Utility grids: (a) conventional grid (b) smart grid. In the conventional system power flows from in one direction only; but for smart grid,
there is no strict structure. Generation can occur at the consumer side too, such as the wind and the solar farms at the outer periphery of the topology.
Power flow can also be bidirectional, demonstrated by the energy storages and the house in this illustration.

Constant connectivity and communication is one of the core systems are required to ensure proper operation of the smart
components of smart grid, and that requires devices equipped grid, safekeeping of data, and thwart any attack aimed at the
with such capabilities. The network created by such devices, power system. Machine learning is an attractive solution
connected to other nodes of the system through the internet, processing big data, and implementing effective security
are called the “internet of things (IoT)”. In the internet of solutions.
things each object has its own identity in the digital world. This paper presents a concise picture of the electricity grid’s
Everything is connected through a complex network. IoT transition towards the smart grid, the ensuing rise in IoT
comprises of smart objects which possess self-awareness, usage, and the challenges this new system brings forward.
interaction with the environment and data processing. Smart The most obvious trials are of course the handling of the
devices are capable to communicate with other such devices huge amount of data in this connected system, their proper
in this system [14]. Most common smart devices employed analysis and safety, as well as protecting this new power grid
in the grid, such as the smart meter, falls into this category. from attacks generated in both physical and cyber
These devices provide the detailed data required for accurate dimensions. This work can act as a base for future academic
information and automated decision support which give the and industrial researchers, while pointing out the current
smart grid the unique capabilities it demonstrates over the limitations with possible solutions along with their
legacy system. All this data need to be handled in real time, effectiveness.
and stored to use historical data to create decisions based on The rest of the paper is organized as follows: a short history
certain cases. Various research works have been conducted of the power grid from its inception is presented in section II.
with data obtained from intelligent devices in substations, Discussion on IoT components, its applications and issues is
feeders, and various databases [15]. Information sources can carried out in section III. Section IV focuses on big data, and
be market data, lighting data, power system data, the role of big data analysis in smart grid. Section V puts
geographical data, weather data etc. [16]. Optimization from forward machine learning as a method capable of handling
generation to distribution requires reliable, accurate and the big data generated in the IoT-based smart grid, and
efficient prediction model for electric energy consumption. highlights its capabilities in renewable energy forecasting.
For example, energy consumption data (kWh) from Emerging security threats to the smart grid, its data, and
100,000’s of customer smart meters at 15 minutes sampling devices are discussed in section VI, including protection and
intervals shows that ensuring the quality of the collected data threat-detection techniques. The excerpt of this detailed study
poses a unique challenge for prediction models and is presented in section VII, with future research directions
evaluation of their efficacy for SG [17]. There are several outlined. Finally, the conclusions are drawn in section VIII.
factors which require to be predicted, such as: renewable
generation, power purchases from energy markets, 24 hour II. CHANGES IN THE CENTURY-OLD GRID
planning of load distribution etc. [18, 19]. These factors are In the early days of electric power systems, AC and DC
the part and parcel of SG sustainability and security [20]. contended to become the industry standard. The AC system
Predictability of electricity consumption has been studied prevailed and have been in use ever since. The reason of
with dynamic demand response in [21]. High volume of data AC’s dominance is its ability to use transformers for
from SG increases the complexity of data analysis. Dynamic changing the voltage level, and enabling the transmission of
energy management (DEM) is required for processing this high voltage electricity which reduces loss. The first
huge amount of data for power flow optimization, system demonstration of the AC transmission system took place in
monitoring, real-time operation, and production planning 1886, at Great Barrington, Massachusetts, USA, by William
[22, 23]. Data of such magnitude, which cannot be processed Stanley and George Westinghouse [29]. Westinghouse later
through traditional processes, is termed as “big data”, and it formed the Westinghouse Electric Company that went on
has also become a centerpiece of current research. supply AC power to Buffalo, New York from the Adams
Researches on big data-based power generation, optimization Power Plant at the Niagra Falls in 1896. Thus the dominance
and forecasting techniques are extended to the renewable of AC system is established, and the worldwide power grid
energy based system such as wind energy system [23, 24]. A adopted this technology as electrification expanded
portion of the data produced in SG contains individual users’ massively over the next century. Now, in the 21st century,
confidential information. This type of data is required to be technology has advanced astronomically as compared to the
protected under legal regulations [25, 26]. Moreover, this late 1800s; however, the grid system in the world largely
data contains classified and sensitive information of an resembles the century-old system that initiated the process of
organization or central grid of a country. Manipulation of electrification. The advanced technologies that emerged in
such data can affect the safe operation of the grid. Therefore the power sector include power electronics, renewable
security and privacy is a very important issue [27]. An IoT energy sources, distributed generation, advanced monitoring
integrated SG is a cyber-physical system [28], which makes and communication system etc. The legacy grid was not
it prone to cyberattacks. Therefore, adequate protection designed to accommodate these devices, and thus they create
significant problems when integrated with the existing grid

Author Name: Preparation of Papers for IEEE Access (February 2017)

infrastructure. For example, power electronics based loads elsewhere physically, and carry out their tasks through the
act differently than the generally perceived loads of resistive, resulting exchange, are termed as IoT devices. IoT stands for
capacitive, and inductive properties; electronics devices “internet of things”, which can be defined as the interrelated
exhibit constant power load (CPL) properties that cause system that links up such devices, and facilitate data transfer
significant system-instabilities [30]. Distributed generation without any human intervention. According to Gubbi et al.
causes bi-directional power flow and thus contradicts the [31], IoT is an interconnection of sensing and actuating
historically unidirectional flow of the grid. Renewable devices providing the ability to share information across
sources often generate intermittent DC power (e.g. solar platforms through an unified framework, developing a
energy) opposing the predominant AC systems. The common operating picture for enabling innovative
renewable sources of AC (e.g. wind energy) are highly applications. This is achieved by seamless ubiquitous
varying as well. All these which makes integrating renewable sensing, data analytics and information representation with
sources in the existing grid a huge challenge. However, even cloud computing as the unifying framework. Each of those
though these next-generation systems disrupt the grid objects has its own embedded computing system which
architecture in place, adopting these technologies is the way enables it to be identified and to be interconnected with each
to move forward, not the other way round. Therefore, the other. The IoT architecture is shown in fig. 2. IoT will consist
current time marks the transition period for the electricity of more than 30 billion objects by 2020 [32]. The astronomic
grid – a metamorphosis that will supplant the archaic system increase in number of IoT devices is visualized in fig. 3.
with an architecture well-capable to accommodate the From a mere 13 billion in 2015, their predicted population
advanced concepts and tools. This next-generation system is reaching 30 billion and beyond in a timespan of five years
called the “smart grid”. The grid evolution timeline in perfectly demonstrates the current trend of IoT application.
presented in Table I. These devices are able to operate with a less amount of
TABLE I external intervention and are capable of responding to the
SIGNIFICANT EVENTS IN THE EVOLUTION OF ELECTRICITY GRID environment spontaneously.
Year Significant events IoT includes, but not limited to, technologies such as
1800 Alessandro Volta made an electric battery to provide a steady medical equipment, smart vehicles, smart grid, smart homes,
supply and smart cities. IoT applications bring forth numerous
1831 Michael Faraday discovered dynamo
benefits. It can reduce human intervention in the process of
interconnecting devices. The most important impacts can be
1878 Invention of incandescent lamp by Thomas Edison (in America)
observed in the power sector, home appliances, and in smart
and Joseph Swan (in Britain)
cities. Smart grids which contain the attributes of IoT may be
1882 DC street lamps appear in New York the possible solution of future global energy crisis. Efficiency
1886 AC system spread rapidly in the following years at transmission and distribution ends can be escalated.
1888 The induction motor got patented independently by Nikola Tesla Renewable energy sources can be more effectively utilized
(in USA) and Galileo Ferraris (in Italy) under IoT based networks. Currently, smart homes have
1896 The Adams Power Plant supplied AC power to Buffalo, New monitoring systems that increases the cost effectiveness [33].
York It also reduces the unwanted consumptions of energy. In a
1976 Power MOSFET appears as a commercial product
smart city, optimization of schedule for public transport can
be done with IoT. However, though the general lifestyle has
2017 Wind power becomes economically feasible
caught up with this technology, it is hardly present in the grid
system. Incorporating these connecting devices in the grid
III. APPLICATION OF INTERNET OF THINGS (IOT) IN
DISTRIBUTED POWER SYSTEM infrastructure is a major step to advance towards smart grid –
The underpinnings those make the smart grid do so many which can be evidenced by the significance put on IoT in
things that the legacy grid is incapable of are a lot of designing microgrids [34]. Niche uses of IoT devices are also
connected devices, which are capable exchanging emerging with applications that are already exists, or
information, and receive commands to act in a certain way. anticipated to appear in near future. Smart homes, where
This extensive communication is made possible by the household appliances can be controlled by connected
internet, and all these devices are connected to their intelligent devices is an example of such use. Connected
respective networks. Devices connected to the internet are vehicles, distributed energy resources (DER), green buildings
currently part and parcels of the daily life, and more and are some more applications [35-37].
more of such devices are emerging every day. An example of
such devices can be smart thermostats. Such devices, which
use the internet to stay connected to resources located

Author Name: Preparation of Papers for IEEE Access (February 2017)

(LPRF) communication using a sub-1 GHz mesh network is

used in USA to convey these data. Wired narrowband
orthogonal frequency-division multiplexing (OFDM)
powerline communication technologies are used in France
and Spain. Energy information can be sent to the demand
side following this same mechanism. IEEE 802.15.4 2.4 GHz
ZigBee® standard is used in USA for this purpose. UK and
Japan are considering sub-1 GHz RF solutions. A combined
implementation with both hybrid RF and powerline
communication can be a feasible way to provide necessary
energy information to consumer homes. Smart meters allow
better tracking of consumption and generation, and better
energy management, among other things.
FIGURE 2. IoT architecture. Data collected by sensors can be sent to Similar to the smart meter, IoT can be integrated in smart
different systems which use various software platforms to carry out grid through all of its major subsystems: generation,
intended tasks [38].
transmission, distribution, and utilization [41-43]. For
example, IoT can provide monitoring services for the power
transmission line, where one part of this monitoring system is
deployed at the transmission line to monitor the condition
and readings of the conductors; another portion of
monitoring system is deployed at the transmission towers.
This portion monitors the environmental conditions of the
towers. A wireless communication technology is used to
establish communication between the transmission line and
the towers. The main monitoring components are listed in
table II [44]. Possible integration of IoT technology in all of
smart grid’s subsystems are listed in table III. The major
security concerns for each of these subsystems are also
mentioned in this table; such security threats are discussed in
detail in section VI.
FIGURE 3. Predicted number of IoT devices over the years. The
astronomic rise in this number demonstrates the recent trend of IoT
application [32, 39].
Smart energy system aims at reducing energy loss while
simultaneously providing sufficient energy and services to
everyone. In India more than 30 percent loss in electricity
occurs during the power production process [40]. In France
and Australia, 35% wastage of water occurs due to the
leakage in the system; therefore, the electricity used in
processing that water also goes to waste. To meet the
increasing demand as well as reduce the energy wastage, a
real-time tracker of supply and demand sides of distribution
system needs to be developed – which IoT can provide.
Centralized systems should be replaced with a distributed
microgrid, which provides real-time monitoring and
communication to the grid, along with remote sensing
technique, two-way communication and demand response.
Useful IoT devices for using in the power sector has
already been developed, smart meter being one of them. The
fundamental concept of smart meters is to provide a two-way
communication simultaneously while measuring power. The
measurement data is transmitted to the utility suppliers
through a mesh network. Low power radio frequency

TABLE II
IOT COMPONENTS FOR MONITORING POWER TRANSMISSION LINES, WITH CORRESPONDING MONITORING TECHNIQUES [44].
Monitoring Item Methodology Usage
Transmission Tower Leaning sensor transmits the status of the transmission tower to the nearby Real time monitoring
Leaning backbone node. Early warning
Conductor Galloping Calculation and analysis of monitoring point can determine the number of Avoiding tower collapse
horizontal and vertical galloping conductors. Possible discharge between phase conductor
Wind deviation It can be calculated by three dimensional accelerated sensors on the finding discharge point
conductors.
Micro-meteorology Wireless sensors can be used. Wireless recording of temperature, humidity, wind
velocity, sunshine, and rainfall
Conductor icing It can be determined by micro-meteorology and tension sensors. Early warning decision making
Alleviating ice flashover
Wind vibration Can be detected with acceleration sensors. Analyzing wind parameters
Conductor Wireless temperature sensor can be used Real time conductor state monitoring
Temperature

TABLE III
IOT APPLICATION IN SMART GRID.
Application Layer IoT application Security Concerns

Generation Monitoring of energy generation Sabotage

Controlling units, gas emission and pollution discharge [45] Data theft
Prediction of power usage [46] Unauthorized use of computational power
Managing distributed power plants and microgrid [41]
Transmission Monitoring and controlling of transmission line and substation Sabotage
Protection of transmission tower [42, 43]
Distribution Distribution automation [47, 48] Sabotage
Management and protection of equipment
Fault management [49]
Consumption Smart homes & appliances [50] Data theft
Intelligent charging and discharging of electric vehicles [51] Identity theft
Power load control
Multi network management

Seamless communication is a core feature of smart grid, demand side. This adjustment of load keeps the other
essential for its proper functioning; and IoT integration can components of grid running. Smart load control and load
aid in smart grid communication too. Mainly four models are shedding should aim at minimizing power outage in sudden
currently being used for communication technologies: device change of a load in the grid. An automated system to do such
to device, device to cloud, device to gateway, and back-end tasks with the help of IoT devices was presented in [58]. This
data sharing pattern [44]. Three layered communication method worked by predicting the day-ahead load, and
systems for IoT implemented smart grid system has also tracking the available generation. When it found the load to
been developed, the layers being: home area network (HAN), be greater than the supply, it could suggest the consumers to
neighbor area network (NAN), and wide area network switch off some appliances, or schedule possible loads to run
(WAN). HAN comprises both wired and wireless at off-peak hours. Fig. 5 shows the working principle of this
technologies, e.g. wired technology is powerline method. Subscriber (consumer) data, weather information,
communications, and wireless communications are ZigBee, and historical data from the grid were used for the prediction
Bluetooth, and WiFi. A home gateway (HG) is a key in this system. All the analysis and decision-making was
component of HAN, which collects data from home conducted in a cloud infrastructure; while the system
appliances. NAN requires a communication system which components communicated through powerline
can cover a radius of more than thousand meters. NAN communication or some wireless technology. Simulation
collects data from the energy meters in HAN and transmits results showed this system to be quicker in responding to
those data to the WAN [52]. WAN serves as interconnection emergencies, and its potential to avoid sudden power outage
between every component of communication link such as [58].
network gateways, NANs, distributed grid devices, utility
control centers, and substations. Core and Backhaul are two
interconnected networks of WAN. Detailed discussions of
HAN, NAN and WAN systems are included later in this
section. Information from the physical systems in IoT
integrated smart grid is fed into data concentrator [53-55].
From data concentrator information is met with the
requirements of internet protocols for web services or cloud
computing platforms. Those web services and cloud
computing platforms further process the data. The energy
service providers’ (ESP) sites are connected with the
Aggregation layer [53-56]. The underlying layers of IoT are
depicted in fig. 4.

FIGURE 5. System architecture for load shedding and smart load

controlling algorithm [58]. All the components of the system are
connected to the cloud which makes decisions based on the system
inputs, and sends out commands for execution. The directions of the
arrows indicate the flow of data.
The integration of IoT devices bring some unique
challenges with them in the smart grid scenario which are
inherent to such technology, and the fact from which all that
stems from is latency. Latency is defined as the difference
FIGURE 4. Structure for IoT implemented layers for SGs [57]. Each IoT between the time of data generation and the time when it
layer corresponds to a certain layer of SG infrastructure. becomes available for applications. In other words, it is the
Efficient load management is a key benefit of employing time delay for the data to become available. Latency in IoT
the IoT technology. As a general practice, system architecture can be characterized as communication latency
disturbances that cause shortage in power generation are and phasor data concentrator (PDC) latency. Communication
compensated by adjusting the amount of load from the latency on the network is comprised of transmission delays,

Author Name: Preparation of Papers for IEEE Access (February 2017)

propagation delays, processing delays, and queuing delays. and wide area measurement system (WAMS) design; and
PDC latency, on the other hand, comprises of PDC device usage of robust analytic functions [62].
latency and PDC wait-time. Wait-time latency indicates the
time each PDC has to wait for certain user-configurable IV. SMART GRID WITH BIG DATA ANALYSIS
time-duration so that slower PMU data can be reached and As it is mentioned in the previous section, integrating IoT
processed for time alignment operation at PDC. The devices in every sector of the grid infrastructure is a
maximum tolerable latency is 40 ms which includes latencies mandatory step for moving towards smart grid. It has also
introduced by communication network, PMU, PDC, etc. This been stated that the defining feature of these devices is their
requirement must be met by any IoT architecture for the ability to communicate with other devices and control
system to function deterministically [59, 60]. centers, and send useful information. Thus, an unprecedented
Communication infrastructure is a critical aspect of IoT amount of data gets generated in an interconnected network
deployment. Typical means of communication are leased [63-66], posing challenges to the conventional methods of
telephone lines, power lines, microwave and fiber optic, data transfer, storage, and analysis. As documented in
among others. IoT functions require real-time processing of reference [67], water consumption data of 61,263 houses in
synchrophasor data at wide area level to aid in making Surrey, Canada amounted to 5 MB, information about speeds
informed decisions. The protection and control commands and locations of vehicles passing through the Madrid
should be available for the destination in a deterministic Highway, Spain generated 450 MB of data, and monitoring a
manner. Communication channel capacity, latency and hence 400 square kilometer area in Cologne, Germany for a day
efficiency, are therefore important for successful created a dataset sized at 4.03 GB – recording information of
implementation of IoT system. Channel capacity defines the around 700 vehicles. Monitoring of transmission line,
amount of data that can be carried by a communication generation unit, substation state, smart metering [68], and
network. In IoT system, all PMUs send phasor data directly data acquisition from smart home - all produce a large
or indirectly to central PDC (phasor data concentrators) for amount of data from the smart grid, which are to be stored in
concentration through time alignment and data aggregation. a cloud-based system for proper analysis. Cloud supported
The aggregated data streams are used by analytic functions to IoT system has been proposed in reference [69] to manage all
identify anomalies and issue corresponding commands to those data.
rectify. With an expected 3000 PMUs transmitting 4 phasors, Enter big data analysis [70], which has become a
6 analog quantities, and 8 digital quantities each to central buzzword in the global scientific and data analyst
PDC, all in floating-point format at 25 or 50 messages per communities [71-73]. Big data refers to massive amount of
second rate, 68 Mbps on a serial port, or 135 Mbps in TCP data that require more advanced methods to be captured,
IP, or 122 Mbps in UDP is required. As discussed in [61], a curated, managed, and analyzed than the traditional tools and
typical serial port cannot handle the above traffic and hence signal processing models. The amount of data that defines
an Ethernet port with TCP/IP or UDP is preferred. big data is not explicitly defined, rather it moves as the
Essentially, the chosen communication infrastructure should technology progresses. Generally, data demonstrating three
enable the bandwidth requirement in a reliable way. characteristics can be labeled as big data: it has a large
Since IoT data contains critical information, it is imperative volume; the velocity or frequency of this data generation,
that the data to be secured from all types of attacks. Attackers storage, or transmission is high; and there is a lot of variation
can modify the data to cause system instabilities or even of data in the dataset. These features match with the data IoT
blackout. To ensure the reliability, a two-layer devices generate, and thus the data generated in the smart
communication security can be constructed: one inside grid can be considered big data. Fig. 6 shows how expanding
substations using already deployed security measures for all in the aforementioned three sectors define big data. Although
data communication, and the other by secured means such as big data means a massive amount of data, technically it
encryption for data stream outside substation. All analytical covers the predictive and behavioral analysis using those
functions using IoT data assume that the incoming data is data. This huge amount of data is available at every aspect of
error free and continuous. But PMU measurements can our lives, and demands critical analysis. Scientists,
become unavailable due to unexpected failure of the PMUs businessmen, social welfare organizations, economists, and
or PDCs or due to loss of communication links caused by many others need to process through this large volume of
congestion of communication network. This missing data information that is available online. Big data analytics is
will result in wrong outputs from the analytical functions. based on this massive data, and the associated analytic
Practical counter measures for reliable and secure data techniques, which is visualized in fig. 7. These techniques
transfer are: building in as much redundancy as possible in are based on different platforms such as Windows, Linux,
PMUs, PDCs and communication; proper PMU placement Mac etc., and they require certain levels of expertise. They

Author Name: Preparation of Papers for IEEE Access (February 2017)

also have certain limitations, which hinders the rise of a

single superior tool. Different tools with their platforms,
required skill levels, and limitations is presented in table IV.
Table V lists some more analysis techniques to juxtapose
their advantages, applications, difficulty level to master,
required system to run, associated software, and financial
cost. From this table, it can be seen that most of the systems
has a high cost involved, and all the high-cost systems have
an ‘expert’ level difficulty. Table VI demonstrates some
notable works on big data analysis in smart grid to point out
their specific applications. Fig. 8 shows the use percentage of
the most-used data science techniques in 2017, where
regression appeared as the most popular one with 60% usage.
Clustering was used by 55% user, while visualization and
decision tree attracted 50% of all users. A 2014 report
published by the National Rural Electric Cooperative
Association (NRECA) of the USA enlisted big data
capabilities as a crucial component of the next generation
power grid, or in other words, smart grid. It states that the
ever-increasing deployment of phasor measurement units
(PMU), and synchrophasors at transmission, distribution, and
distributed generation sectors will generate massive amounts
of data - which will vary as the direction of power flow will
change depending on seasonal and daily conditions [74].
Such deployments of PMU can also lead to proactive control FIGURE 6. Big data characteristics: large volume of data with lots of
of grids, preventing faults from taking place instead of variations which are generated, stored, or transmitted at a high at a high
velocity can be labeled as big data.
clearing a fault after its occurrence [75]. Analyzing big data
is stated as a key functionality for energy management
systems (EMS) for smart grids, control algorithms, and
future energy market models in references [76-83]. Zhou et
al. presented ways to determine residential energy
consumption through big data analysis in reference [84].
Demand side management through bid data analysis has also
caught much attention [85, 86]. Dynamic energy
management through such data analysis is also a promising
technology [23].

FIGURE 7. The components that create big data analytics. Big data and
the techniques to analyze it has created the discipline of big data
analytics [87].

TABLE IV
DIFFERENT ANALYTICAL TOOLS FOR BIG DATA, THEIR PLATFORMS, REQUIRED SKILL LEVELS, AND LIMITATIONS (ADAPTED FROM [88]).
Tool Category Platform Skill Level Limitations

Data Data cleaning Browser Advanced begineer Sends data to an external site which
Wrangler compromises securtiy.
R Project Statistical analysis Linux, Mac, Unix, Intermediate to advanced Takes time to adapt; interface is text-only;
Windows has limited memory.
TimeFLow Temporal data analysis Desktop+Java Beginner No option available for exporting results.
NodeXL Network analysis Windows Expert Problems with application program interface
(API).
CSVKit CSV (comma-separated values) Linux or Mac with Expert Slow to adapt; Dependant on Python.
file analysis Python
Tableau Visualization app Windows Advanced beginner to Sends data to public website which
intermediate compromises securtiy.

TABLE V
ADVANTAGES, APPLICATION, DIFFICULTY LEVEL, SYSTEM REQUIREMENT, SOFTWARE PLATFORM, AND COST ASSOCIATED WITH SOME BIG DATA
ANALYSIS TECHNIQUES (ADAPTED FROM [89]).
List of Methods Advantage Application Difficulty System Required Software Cost
Level Involved involved
Machine learning Strategic resource usage, Building and training Expert Training dataset, AI Python, Matlab High
simplified management. intelligent system. algorithms.
Data mining Summarizing relevant Data extraction, data Expert Crawler, NLP. Python, R, High
information from a large cleaning, data labeling. Matlab
amount of data.
Genetic Saving time during traning, Training optimization. Expert Genetic representation, Matlab, Python High
llgorithms quick convergance. fitness function.
Neural networks Less mean squared error Anomaly detection, Expert Training dataset, Matlab, Python High
(MSE). pattern recognition, hidden layer,
prediction. optimization.
Natural language Handling text data. Analyzing and Expert NLP tool Python High
processing (NLP) visualizing text data.
A/B testing Figuring out the best Web analytics. Beginner Browers Google Medium
strategies. analytics,
Optimizely
Cluster Gaining insights from data. Data grouping, Intermediate Classifers Python, Matlab Medium
classification.
Crowd-sourcing Human intuition, real time Gathering large scale Beginner Web page FeatureHub Low
analysis. data features.

service provider domain, smart appliances provide

TABLE VI information for energy price estimation [23, 103]. A flask
APPLICATIONS OF BIG DATA ANALYSIS IN SMART GRID.
Reference Institute Year Application
framework called OASIS dashboard was developed for the
Zhou et al. Hefei University of 2016 Determining visualization of real time energy data from the energy
[84] Technology, China residential energy sources in Puerto Rico in [96]. Reference architecture for
consumption smart grid using big data and intelligent agent technologies
Zhou et al. Hefei University of 2015 Demand side
[85] Technology, China management was developed in [104]. In this architecture, agent-oriented
Zhou et al. Hefei University of 2016 Demand side programming methodologies were adopted in Hadoop
[86] Technology, China management platforms [104]. This system supported interoperability in
He et al. Shanghai Jiaotong 2017 High-dimension
[90] University, China (with smart grid
smart grid systems. Application of random matrix theory
external collaboration) modeling (RMT) for high-dimension smart grid modeling was
Ryu et al. Sogang University, Korea 2016 Short time load side demonstrated in [90], which claimed to provide better
[91] (with external prediction accuracy, and practicality for large interconnected systems
collaboration)
Coelho et State University of Rio de 2017 Load forecasting such as smart grid. Five case studies conducted in that work
al. [92] Janeiro, Brazil (with verified the proposed system’s capabilities.
external collaboration) For the development of energy efficient and sustainable
Bessa et al. INESC Technology and 2015 Very short-term
data processing system, a framework with robust time
[93] Science, Portugal (with solar power
external collaboration) forecast advanced workload and energy management was developed
in reference [105] to integrate renewable energy sources,
distributed storage unit, dynamic pricing unit etc. for green
DC systems. In that work, a resource allocation problem was
developed so that the net cost of the system could be
minimized with spatio-temporal variations of workloads and
electricity market prices. Net cost of the system comprised of
network operational cost and the worst-case energy
transaction cost. An optimal solution was achieved by
Lagrange dual based distributed solver using strong duality
of convex reformulation [105]. Large amount of data from a
power system require fast and efficient computing, which has
been a concern for several researchers. Task parallelism with
multi-core, cluster, and grid computing can reduce the
computational time in an efficient data mining algorithm
[106]. A grid computing framework was developed for
higher computing efficiency in [107]. In this framework, the
overall architecture consisted of three layers: resource layer,
grid middleware and application layer [107]. The data
generated in the smat grid raises two major concerns.
Firstly,the data must be processed and transferred in an
efficient way within an acceptable limit of time. And
secondly, security concerns are very important issues
regarding IoT integrated smart grid [108]. To provide an
insight on these issues, the next two sections are organized to
FIGURE 8. Most used data science techniques in 2017. Regression tops
this list, attracting 60% of the users [94].
address these concerns.
Agelidis et al. mentioned big data as one of the challenges in
V. MACHINE LEARNING APPLICATION IN SMART GRID
the information and communication technology part of smart
The obvious question that arises from the big data generation
grid [95]. Data in smart grid come from various sources.
from smart grid is efficient ways to analyze them for
Mainly two domains provide smart grid data: generation
extracting valuable information. Without the extraction of
domain and service provider domain [96]. Uncertainty in
useful information, the collected data holds little or no value.
data analysis demands development of methods for long and
Machine learning appears as the tool required for the tall task
short-term data patterns from distributed energy resources
of going through the massive amount of data generated in an
(DER). Few of the proposed models are Gaussian model
IoT-based grid system. It fits in as the final piece of the smart
[97], finite state Markov models [98], hidden Markov model
grid system which is driven by data collection, analysis, and
(HMM) [99, 100], proportional integral derivative (PID)
decision making. Machine learning techniques provide an
controller [101], and online learning techniques [102]. At the
efficient way to analyze, and then make appropriate

Author Name: Preparation of Papers for IEEE Access (February 2017)

decisions to run the grid; and thus enables the smart grid to management employing entropy based feature selection,
function as it is intended to. machine learning, and soft computing was proposed by
Machine learning (ML) is a term which refers to learning Jurado et al. in [132]. Several algorithms such as extreme
and making predictions from available data by a system. It is learning machine, support vector regression, improved
comprised of various algorithms which analyze the available second order, decay radial basis function neural network, and
data through a set of instructions to produce data-driven error correction to train common radial basis function
predictions and/or decisions. Machine learning undergoes the networks for predicting load was investigated in [133]. Ryu
rigorous process of designing and programming explicit et al. proposed a deep neural network (DNN) load
algorithms with expected performance. The steps and forecasting method for short term prediction at the load side
associated standards of this process are presented in table which demonstrated as high as 29% less error compared to
VII. Machine learning functionalities include predictions of existing systems such as shallow neural network (SNN), and
consumption, price, power generation, future optimum double seasonal Holt-Winters (DSHW) [91]. A graphics
schedule, fault detection, adaptive control, sizing, and processing unit (GPU) based load forecasting method has
detection of network intruders during a data breach [109- been proposed in [92]. A 45 MW smart grid in University of
116]. Xu et al. presented an assessment model for analyzing California San Diego (UCSD) is considered in [134]. This
transient stability which employed extreme learning grid supplies 54000 consumers from both renewable and
machine, and demonstrated impressive accuracy and non-renewable energy sources. The UCSD grid is equipped
computational speed when tested on New England 39 bus with advanced monitoring and storage techniques. In that
system [117]. Wang et al. pursued a similar objective with work, big data analyses have been done leveraging large
their novel core vector machine (CVM) algorithm to utilize amount of data and the Hadoop system. Machine learning
big data generated by PMU, their system was also tested on can also be applied for various security applications in smart
the New England bus system [118]. For transmission grid. A concise presentation of such uses is shown in fig. 9.
systems, machine learning can be employed to analyzed the However, the most promising and much needed use of
phasor measurement unit (PMU), and micro phasor machine learning in the next generation energy system is the
measurement unit (µPMU) data for uses such as system renewable energy sector. And therefore, in the following
visualization and frequency detection. Machine learning can subsections, implementation of machine learning in SG with
be used in these purposes alongside other software such as renewable energy sources is discussed.
power plant model validation tool (PPMV), and free flight
risk assessment tool (FRAT). Several machine learning
methods are being introduced at different phases of
renewable energy power system based SGs, creating a whole
new prospect for research [119-121]. For example, the
support vector machines (SVM) have been widely
implemented into several problems of renewable energy
power systems, which provided many optimization and
prediction techniques in SG [122-124]. Economic
optimization for smart grid prosumer node with a two-level
control scheme is developed in [125]. Machine learning
based fast and accurate algorithm for monitoring power
quality (PQ) events in an SG has been developed recently in
[126, 127]. Li et al. applied machine learning to analyze user
predilections in a smart grid to find out usage pattern and
preferences [128]. In [129], Remani T. et al. demonstrated a
generalized use of reinforced learning to schedule residential
load considering renewable energy sources and all possible
tariff types. For distributed generation systems, islanding
detection using machine learning and wavelet design was
investigated in [130]. Application of particle swarm
optimization (PSO) to enhance stability for unplanned
islanding in microgrid is proposed in [131]. Big data analysis
to monitor and detect such islanding incidents comes before
this stabilization stage. A hybrid system for demand side

TABLE VII
STEPS INVOLVED IN MACHINE LEARNING AND DATA MINING [135].
Steps involved Standards

Fayyad Cios SEMMA CRISP-DM

(Sample, Explore, Modify, Model, Assess) (Cross Industry Standard Process for Data Mining)

Determining objective ◼ ◼ ◼

Collecting data ◼ ◼ ◼ ◼

Cleansing data ◼ ◼ ◼ ◼

Reducing data ◼ ◼ ◼

Reformulating problem ◼

Explorating data ◼ ◼

Selecting tools ◼ ◼

Constructing model ◼ ◼ ◼ ◼

Validating model ◼ ◼ ◼ ◼

Interpretating result ◼ ◼ ◼ ◼

Deployment ◼ ◼ ◼

[158]. Machine learning technique has been applied to

diagnose wind turbine faults using operational data from
supervisory control and data acquisition (SCADA) system of
south-east Ireland [159]. Generalized mapping regressor
(GMR) was employed in reference [160] to create steady-
state model of wind farms that can help in detecting faults in
case of an anomaly. Fan et al. employed Bayesian clustering
to create a dual stage hybrid forecasting model to aid in
scheduling of a wind farm, and trading of wind power. This
proposed system was validated by applying on a 74 MW
wind farm at Oklahoma, United States [161]. Parallel
execution of Guassian process and neural network sub
models to predict wind power was presented in [162]. Short
term wind power prediction using ELM and coral reefs
optimization (CRO) algorithm was presented in [163] which
demonstrated superior performance when applied to a wind
farm in the United States. A similar objective was pursued in
China through hybrid machine learning models based on
variational mode decomposition and quantile regression
averaging, which attained absolute error as low as 4.34%
[164]. Improved simplified swarm optimization (iSSO), an
improvement of simplified swarm optimization by means of
FIGURE 9. Application of machine learning in smart grid security. bias and weight justification, showed impressive results
Unsupervised and supervised – both approaches can be used to carry
out an array of tasks including threat identification and data when used to predict wind power generation at the Mai Lao
categorization [136]. Wind Farm at Taiwan [165]. Hybrid swarm technique (HAP)
consisting of particle swarm optimization (PSO) and ant
A. MACHINE LEARNING APPLICATION IN WIND colony optimization (ACO) to predict wind power in short
ENERGY FORECASTING
term from parameters such as ambient temperature and wind
Wind power is one of the fastest growing renewable energy
speed is presented in [166]. This system achieved a mean
sources in the world. About 12% of the world’s electricity
absolute percentage error rating of 3.5%. Wang et. al.
will be supplied by wind generation by 2020 [137].
developed a novel hybrid strategy based on a three-phase
Integration of wind power sources with the grid provides
signal decomposition (TPSD) technique, feature extraction
several technical, economic, and environmental benefits
(FE) and weighted regularized extreme learning machine
[138]. But due to the intermittent and stochastic nature of
(WRELM) model in [167]. This model was able to do a
wind power, it provides some obstacles during power
multi-step ahead wind speed prediction. In this model, a
generation and distribution. Variation in wind speed causes
three-phase signal decomposition framework including
fluctuation in the output of wind power plant, which leads to
seasonal separation algorithm (SSA), fast ensemble empirical
instability in the grid. Hence proper forecasting is required
mode decomposition (FEEMD), and variation mode
for wind energy based power grids, and can aid in making
decomposition (VMD) were used to control the unstable and
operational strategies [139-142]. This forecasting is complex,
irregular natures of wind speed. An FE process including
because from controlling wind turbine to integrating wind
partial autocorrelation function (PACF) and regression
power into energy system, time duration for prediction
analysis was used to utilize the effective and beneficial
changes from milliseconds or seconds to minutes or hours
features of wind speed fluctuations. In this way, the optimal
[143, 144]. Previously, several prediction models such as
input features for a prediction model was determined. To
fuzzy modeling [145], auto regressive moving average [146],
improve the forecasting accuracy and efficiency, an
artificial neural network [147, 148], K-nearest neighbor
improved extreme learning machine (ELM), named weighted
classification [149], computational fluid dynamics (CFD)
regularized extreme learning machine (WRELM) was
pre-calculated flow fields [150], extreme learning machine
developed by utilizing these features. Application of
(ELM) [151-153], adaptive neuro-fuzzy inference system
reinforcement learning in energy trading in smart grids with
(ANFIS) [154], combination of relevance vector machine
wind energy generation was demonstrated by Xiao et al. in
(RVM) and differential empirical mode decomposition
[168]. Their proposed system used historical energy trading
(EMD) [155], combination of soft computing model and
data, and energy price, to reduce power plant scheduling. The
wavelet transformation [156], wavelet transform and SVM
energy exchange scenarios between microgrids were also
[157] etc. have been developed and applied. Applications of
investigated using game-theory approach in [169], where it
data mining for prediction of wind power was reviewed in

Author Name: Preparation of Papers for IEEE Access (February 2017)

was shown that overstated trading information can result in for such prediction in [185]. This method outperformed
reduced utility of smart grids. statistical regression algorithms in terms of robustness to
prediction numbers, bias, and accuracy. A hierarchical model
B. MACHINE LEARNING APPLICATION IN SOLAR was proposed based on the machine learning algorithms by
ENERGY FORECASTING Li et. al. in reference [186]. In this work, 15-minute averaged
Solar energy is one of the most prominent renewable energy power measurements were collected from the year 2014.
sources. Solar photovoltaic (PV) systems had 22 GW of Computing error statistics models were used to test its
global capacity in 2009 and almost 139 GW in 2013 [170, accuracy. The hierarchical forecasting approach utilized
171]. Similar to wind energy sources, the solar power machine learning tools at a micro level to predict each
systems too are impeded by many difficulties. Many natural inverter performance. Then it evaluated the performances at
and man-made impediments such as weather conditions, a larger level by adding up the micro level predictions. In
seasonal changes, topographic elevation, discontinuous that way, it provided a bigger picture of the plant. This
production, and intra‑hour variability have effects on solar framework is visualized in fig. 36. Table VIII summarizes
PV system performance. As a result, solar energy the applications of machine learning techniques in renewable
information should be acquired in advance to minimize the source integrated smart grid encountered in literature.
operating costs caused by the various obstacles mentioned
above. Prediction models for both meteorological forecasts
and system outputs were presented in references [172-178].
A forecasting approach aiming at very short-term solar
power forecast based on the city of Évora, Portugal was
proposed in reference [93]. This model used a vector
autoregressive model fitted with recursive least squares. Data
from smart meter and other smart components at medium
voltage/low voltage (MV/LV) substation level were used in
this model. Chaouachi et al. proposed a neuro-fuzzy system
for maximum power point tracking (MPPT) in 20 kW solar
photovoltaic (PV) system in [179]. This method utilized
classifier running on fuzzy logic in accordance with three
artificial neural network having multiple layers. Reference
[180] utilized those two methods for an intelligent energy
management system that could predict PV generation 24
hours ahead. Another day-ahead forecasting method that
could take weather data into consideration was presented in
[181] by Yang et al. Their method was a hybrid one,
employing three different machine learning techniques in the
three stages of the prediction system, and was trained on data
collected from the Taiwan Central Weather Bureau. Hossain
et al. emlpoyed machine learning techniques such as
multilayer perceptron (MLP), least median square (LMS),
and support vector machine (SVM) in forecasting of solar
power in two-phase experiment, where the second phase
concentrated on parameter optimization to find out
performance enhancement margin before and after such
optimization [182]. They concluded that increased attention
in parameter optimization and selection of feature subset
could go a long way to increase prediction accuracy. Li et al.
proposed a solar irradiance forecasting technique employing
SVM regression and hidden Markov model in [183]. A coral
reefs optimization - extreme learning machine (CRO–ELM)
algorithm was proposed in reference [184] to predict solar
irradiation worldwide – which demonstrated better
performance than conventional ELM and SVM. Salcedo-
Sanz et al. also employed Gaussian process regression (GPR)

TABLE VIII
APPLICATION OF MACHINE LEARNING TECHNIQUES IN SMART GRIDS WITH RENEWABLE ENERGY SOURCES.
Reference Institute Year Machine Technique Application
learning strategy
Xu et al. Hong Kong Polytechnic 2011 Data Extreme learning machine Assessment model for
[117] University, Hong Kong (with categorization analyzing transient
external collaboration) stability
Wang et al. Wuhan University, China 2016 Classification Core vector machine (CVM) Analyzing transient
[118] (with external collaboration) stability
Li et al. [128] The University of Oklahoma, 2011 Pattern Machine learning Finding out customer
USA recognition usage pattern and
preferences
Alshareef et University of Ontario Institute 2014 Detection & Machine learning and wavelet design Islanding detection in
al. [130] of Technology, Canada Classification distributed generation
systems
Jiang et al. National Renewable Energy 2017 Optimization Particle swarm optimization Enhancing stability for
[131] Laboratory, USA (with unplanned islanding in
external collaboration) microgrid
Jurado et al. Sensing & Control Systems, 2015 Data Entropy based feature selection, machine Demand side
[132] Spain (with external categorization & learning, and soft computing management
collaboration) Optimization
Marvuglia et CRP Henri Tudor/CRTE, 2012 Detection Generalized mapping regressor (GMR) Detecting faults in wind
al. [160] Luxembourg (with external farms
collaboration)
Fan et al. Monash University, Australia 2009 Classification & Bayesian clustering Scheduling wind farm and
[161] (with external collaboration) Optimization trading of wind power
Lee et al. University of Texas at Austin, 2014 Prediction Guassian process and neural network Predicting wind power
[162] USA
Salcedo-Sanz Universidad de Alcalá, Spain 2014 Prediction & Extreme learning machine and coral reefs Short term wind power
et al. [163] (with external collaboration) Optimization optimization (CRO) algorithm prediction
Zhang et al. Wuhan University, China 2016 Data Variational mode decomposition and quantile Short term wind power
[164] (with external collaboration) categorization & regression averaging prediction
Prediction
Yeh et al. University of Technology 2014 Prediction & Improved simplified swarm optimization Predicting wind power
[165] Sydney, Australia (with Optimization (iSSO) generation
external collaboration)
Rahmani et Universiti Teknologi Malaysia, 2013 Prediction & Particle swarm optimization (PSO) and ant Short term wind power
al. [166] Malaysia (with external Optimization colony optimization (ACO) prediction
collaboration)
Wang et. al. Nanjing University of 2018 Prediction Hybrid strategy based on a three-phase signal Wind speed prediction
[167] Information Science and decomposition (TPSD) technique, feature
Technology, China (with extraction (FE) and weighted regularized
external collaboration) extreme learning machine (WRELM)
Chaouachi et Tokyo University of 2010 Detection Neuro-fuzzy system Maximum power point
al. [179] Agriculture and Technology, tracking (MPPT) in solar
Japan photovoltaic (PV) system
Chaouachi et Tokyo University of 2013 Prediction Neuro-fuzzy system Intelligent energy
al. [180] Agriculture and Technology, management system for
Japan (with external photovoltaic generation
collaboration) prediction
Yang et al. National Cheng Kung 2014 Prediction Hybrid machine learning employing learning Forecasting solar
[181] University, Taiwan (with vector quantization (LVQ), self-organizing generation considering
external collaboration) map (SOM) network, and Support vector weather data
regression (SVR) at different stages
Hossain et al. Central Queensland 2013 Prediction Multilayer perceptron (MLP), least median Forecasting solar power
[182] University, Australia square (LMS), and support vector machine
(SVM)
Li et al. [183] CSIRO CCI, Australia (with 2016 Prediction SVM regression and hidden Markov model Forecasting solar
collaboration) irradiance
Salcedo-Sanz Universidad de Alcalá, Spain 2014 Prediction Coral reefs optimization – extreme learning Predicting solar
et al. [184] (with external collaboration) machine (CRO–ELM) algorithm irradiation worldwide
Salcedo-Sanz Universidad de Alcalá, Spain 2014 Prediction Gaussian process regression (GPR) Predicting solar irradiance
et al. [185] (with external collaboration) worldwide

VI. CYBER SECURITY IN SMART GRID

A. CYBER-SECURITY CHALLENGES IN SMART GRID

As IoT integrated smart grid systems create a complex
interconnected web as well as a large volume of data which
is often stored in cloud storages, breach of data security is a
serious concern [187]. Threat to the security of this complex
network is always very critical and sensitive as both the
demand side and the supply side of power system are
affected [188, 189]. Various types of security threats for IoT
integrated SG system is depicted in fig. 10 [190]. Such
cyber-attacks to smart grid systems can be carried out to
cause damage to its crucial components, to gain foothold or
superiority in its control system for exploitation, economic
intimidation, or sabotage. The system-assets targeted in-
general in cyber-attacks are depicted in fig. 11, which shows
financial information and intellectual property being the most FIGURE 10. Security concerns of IoT integrated SG system. These can
be categorized into four major types: physical threat, network threat,
attractive targets to attackers, attacked almost 80% of the software threat, and encryption threat. The security concerns under
instances. Research data and other information as well as each type are marked in corresponding color.

control systems are the other major targets. Information

generated from the IoT devices comprise of most of the
targets, and that alone can suffice to demonstrate the
magnitude of the need of cyber-security for smart grid
systems. Table IX shows the information that are available in
a smart grid, and can be targeted for attacks. The components
vulnerable to cyber-threats in a utility’s digital infrastructure
are shown in table X. The vulnerability of IoT devices to
cyber-attacks rated them the most probable way of attacking
both in 2016 and 2017 [191, 192]. Potential sources of
attacks over the years are shown in fig. 12. Along with IoT
vulnerabilities, data breaches at corporate officer levels were
the second most serious threat in 2016 (shown as “CxO
issues”). 2017 saw the rise of automation vulnerabilities,
along with malware and ransomware attacks – quite a few of
which were state-sponsored. The authors predict the use of
artificial intelligence to be the most serious weapon in cyber-
FIGURE 11. Typical targets of cyber-attacks: intellectual property and
attacks, with advanced malwares, ransomwares, state- financial information are the two most sought after assets for attackers.
sponsored attacks, and weakness in data protection
regulations. The most probable outcomes of cyber-attacks to
smart grids can be: operational failures, synchronization loss,
power supply interruption, high financial damages, social
welfare damages, data theft, cascading failures, and complete
blackouts [193]. Direct impacts of blackouts can be
production loss and business shutdown, food spoilage,
damage of electrical and electronic devices, data loss,
inoperability of life-support systems in hospitals and
elsewhere, loss of critical infrastructure such as waste-water
treatment plants etc. Indirectly, blackouts may result in
property loss from arson and looting – which was observed
in many previous occasions, overtime payment of personnel
engaged in emergency management, potential increase of
insurance rates etc. [194].

TABLE IX
AVAILABLE INFORMATION IN A SMART GRID SYSTEM THAT CAN BE TARGETED IN CYBER-ATTACKS [195].
Data Element Type of Asset Description

Name Subscriber Party responsible for the account

information
Address Subscriber Location where service is being provided
information
Account number Financial Unique identifier for the account
information
Meter reading Internal control KWh energy consumption recorded at 15-60 minutes interval during the existing billing cycle
Current bill Financial Amount due on the account
information
Billing history Financial Past meter bills including history of late payments
information
Home area Internal control Network information of in-home elctrical appliances and devices
network
Lifestyle Subscriber When the home is occupied and unoccupied, when occupants are awake and asleep, usage history of different
information appliances
Distributed Intellectual property The presence of on-site generation and storage devices, operational status, net supply, consumption from the
resorces grid, usage patterns
Meter IP Internal control The internet protocol address for the meter, if applicable

units (PMU) and control center etc. [196, 200]. FDIAs are
TABLE X modeled mathematically in [197, 201]. The advanced
VULNERABLE COMPONENTS OF DIGITAL ELECTICAL UTILITY
INFRASTRUCTURE THAT CAN BE TARGETED FOR CYBER-ATTACKS [194]. metering infrastructure (AMI) is one of the most targeted
Component Related Assets parts of a smart grid for cyber-attacks due to its large
proportions and cyber-physical properties [202]. Wei et al.
Billing and Debt Collection (BDC) Customers, personal
information
listed the attacks on AMI components such as smart meters,
Supervisory Control and Data Acquisition National Utility Network, and communication network in [202], FDIA, and distributed
(SCADA) Centralized system denial of service (DDoS) appeared as risks in both sectors.
Programmable Logic Controller (PLC) Subsystem of SCADA, Fault Thus false data injection poses a significant threat that can be
Recorders
Energy Management Systems Electric Utility Grid
carried out in stealth to misguide the state estimation process,
(EMS)/Distribution Management Systems information and disrupt measurement and monitoring systems in smart
(DMS) grid [199]. Energy trading is also a prominent feature of
Cyber Physical System (CPS) Overall monitoring smart grids, which requires the exchange of energy prices,
elements, Computer based
algorithms contracts, and transactions between grid entities. Because of
Power Line Communication Smart Meters, Public these, the system attracts attacks including availability
Cellular Network attacks, integrity attacks, and confidentiality attacks. If the
energy trading sector is exploited, energy, money, and data
theft as well as DoS attacks are possible [203].
Among consumer-level energy appliances, home energy
management system (HEMS) is a common one. In an
HEMS, security and privacy of communication infrastructure
is provided by the home gateway (HG) system. Network and
software attacks are both capable of damaging HEMS. Smart
meters which record energy data from the user-end to billing
have a connection which is stable and trustable with the
home gateway system. Neither of these devices can be shut
down remotely. They do not have the physical access of
HEMS. Smart meters use wireless LAN (WLAN) and other
communication networks which should be tamper proof.
Home gateway serves as a communication channel of SG. Its
configuration is controlled by the suppliers. Any error in the
data can be reported to the meter point operator (MPO).
FIGURE 12. Predicted attack sources over the years. IoT always
Network attack is the most important concern of HG. Every
remains a major vulnerability against cyber-threats. “CxO issues” component of HG has cryptographic key-stores, which use
indicate data breaches at corporate officer levels.
different protocols for secret key generation, key exchange
In a data-based system like the smart grid, false data and management. Different protection levels are associated
injection can have devastating effects, and that motivation with each protocol. Table XI compares some of the most
acts behind false data injection attacks (FDIA). The objective common attack types to demonstrate their relative effects on
of such attacks is to alter original data in an attempt to smart meter systems, and associated financial impacts. It can
mislead the system. Load distribution attack, stealthy be concluded from this comparison that availability attacks
deception attack, covert cyber deception attack, data integrity are the most severe ones, as they have the most adverse
attack, and malicious data attack – all these terms are also effects on the smart meter systems, while all three attack
used to mention such attacks [196, 197]. FDIAs need to be types have serious financial tolls. If compared within the
capable of escaping bad data detection (BDD) protocols in availability attack categories, radio frequency jamming, and
place, and perform stealth attacks on the system state reply attacks are the most effective ones financial sabotage
estimation mechanism [196] – which is fundamental to and smart meter communication blockade; however, denial
monitor the state of a power system [197]. Also, most of the of service (DoS) is the weapon of choice for inducing delay
legacy BDD systems fail to detect such attacks [197]. Along in the smart meter system effectively [204].
with affecting the state estimations, FDIA can disrupt
electricity markets through false economic dispatch and data
[196, 198, 199]. False data can occur in the cyberspace, or in
the physical space to affect device operation. These can
result in flooding of a communication network, corruption of
data, authentication failure, replacement of data packet from
communication channels connecting phasor measurement

Author Name: Preparation of Papers for IEEE Access (February 2017)

N N!
TABLE XI Total  ( )  (6)
COMMON ATTACK TYPES AND THEIR IMPACTS (ADAPTED FROM [204]). k k !( N  k )!
Attack Financial Delay in Duration of Now if N-2 contingency is considered (taking k = 2), the
Impact Smart Smart Meter
Meter Communication
possible combinations of simultaneous attacks will be:
System Blockade
Denial of * ** * N N! N ( N  1) (N 2  N )
service ( )  ( ) (7)
Radio *** * *** 2 2!( N  2)! 2 2
Availability
frequency Therefore, it is very much possible to cripple power
attacks
jamming
Reply *** * ***
systems with well-planned attacks.
attacks Cyber-attacks on infrastructure is a very possible reality. In
Integrity attacks *** * * the fiscal year of 2014 alone, 79 such attacks on the United
Confidentiality attacks *** * * States energy companies have been recorded by a
***: extreme, **: moderate, *: mild Department of Homeland Security division. In 2013, this
number was 145. 37% of USA energy companies failed to
To visualize the process of cyber-attacks in a smart grid prevent attackers in the time-period of April 2013 to 2014
system, a simple scenario can be considered where the state [208]. During 1994-2004, the transmission system was
of a power system is expressed in complex magnitudes of attacked the most worldwide – a staggering 62% of all
voltages and bus angles. Taking the voltage magnitudes as V, attacks in this period were aimed at this part of the power
and the angles as  , the state vector as S can be defined as system, as presented by the Journal of Energy Security [209].
[193]: Attack percentages on all major power grid components in
this time are presented in fig. 13. Some recent attacks are
S  [1 2 3 ...... nV1V2V3 ......Vn ]T (1) presented in table XII. In this age of connected systems,
The state estimation can be stated as below [205]: cyber-security thus appeared as a serious concern in the
energy sector. 63% of utility executives believe their
m countries’ utility grids face significant of moderate risks of
min J ( X )   wi ( zi  hi ( X )) 2 (2) being targets of cyber-attacks in the next five years, as found
i 1 in a global survey conducted in October 2017. Their
Here, h(X) acts as the measurement function to represent concerns regarding electric supply interruption from cyber-
the measurement of the weights: z and w, m is the maximum attacks are visualized in fig. 14, which shows moderate
number of measurement. Without any error, likelihood of attacks worldwide is almost 50%, with a similar
scenario over North America, Asia Pacific, and Europe.
Zi  hi ( X ) (3) However, while considering significant likelihood, Europe
With error, this will be: expects the least amount of attacks [210]. A survey
conducted by McAfee in 2007 documented large-scale
Zi  hi ( X )  ei (4) DDoS attacks. Frequencies of those attacks on infrastructures
of different countries are shown in fig. 15. Brazil’s systems
Here e indicates the measurement error. Now if a cyber-
appear to be the most attacked ones, hit 80% of the time,
attack aimed at inserting malicious data in this power system
followed closely by India, France, Spain, and Italy [209].
is launched, and it succeeds in modifying the measurement
This survey contrasts with the one presented in fig. 16, as the
data with an attack vector,  , then the control system will
most three of the most hit countries (France, Spain, and Italy)
receive the following measurement data:
are European. But these attack statistics are from 2007, and
the survey visualized in fig. 15 is from 2017: which
Zi  hi ( X )  ei   (5) demonstrates the significant improvement in European
For contingency analysis [206], with W&W 6 bus system cyber-infrastructure that almost negated significant
considered as the benchmark, power security was intended to likelihood of cyber-attacks in that region. Both of these
be maintained for N-1 contingencies by the North American surveys, however distant their time periods are, placed the
Reliability Corporation (NERC) [207]. Even so, power United States as a prime target of attacks. Then it is no
systems remain exposed to damages resulting from outages surprise to find this country as the one facing the most
in multiple branches – for example N-k contingencies. For N damages – $17.6 million – as documented in a 2017 report
number of branches, total contingencies to be considered for [211]. These losses faced by different countries are presented
k outages can be formulated as: in fig. 15, where no other country faced so much penalty as

Author Name: Preparation of Papers for IEEE Access (February 2017)

the US, and Australia was the least hurt – capping the
damage costs at $4.3 million. But being the target of the
majority of cyber-attacks may have made the United States
evolve as one the most prepared countries to face such
adversaries, as shown in fig. 17 [212]. However, this 2015
preparedness index did not help the city of Atlanta in the
state of Georgia, USA, when most of its municipal activities
shuddered to a halt after being attacked by a ransomware on
March, 2018. This situation persisted for five days, after
which the system recovered partially [213]. This is just a
demonstration of the fast-evolving cyber-threats, and the
need of better counter-measures.

FIGURE 13. Attacks on major power grid components during 1994-

2004; the transmission system faced most of the attacks, reaching 62%.

TABLE XII
SOME RECENT CYBER-ATTACK INCIDENTS.
References Year Location Attacked System Impact of the Attack

[214] September, Siemens systems at U.K., Windows operation system Unstable power system operation.
2010 North America Korea, and
Iran

[215] August, Midwest and Northwest Software program of the Up to 4-day-long blackouts in some parts, affecting around
2003 U.S and Ontario, Canada cyber system 50,000,000 people and 61,800 MW of electric load.
[215] September, Italy and Switzerland Communication system Disruption in power supply affecting a total of 56,000,000
2003 within the power grid people. 18 hours of blackout in Italy causing massive financial
operators damage.
[215] November, Southwest Europe Communication system Large blackout.
2006
[216] December, Ukrainian Kyivoblenergo Computer and SCADA Blackout lasting 3 hours, affecting 225,000 people.
2015 System
[217] June, Ukraine Network system Power providers, major banks, government and airport
2017 computers taken out of service.
[213] March, Atlanta, Georgia Municipal system City government’s computer systems, traffic ticket system,
2018 water bill payment system, and airport WiFi were taken out by
ransomware, affecting around 6,000,000 people.

FIGURE 14. Likelihood of electric supply interruption from cyber-

attacks, as predicted by utility executives. Moderate likelihood of such
attacks are almost the same globally, but significant likelihood of
attacks on European utilities is very low. FIGURE 17. The countries most prepared to fend off cyber-attacks. All
these countries have a very similar preparedness profile, though being
leaded by the United States by a very small margin.

B. COUNTER-MEASURES FOR CYBER-THREATS

Deploying smart grid system has far reaching impacts on an
organization, and affects all the components of its
technological infrastructure. Thus, security measures also
need to be equally pervasive. Cyber-security strategies can be
divided into two primary categories: protection and
detection. Protection strategies can be hardware and
administrative levels alongside the most-obvious software
safeguards, while the detection can be done by applying
machine learning techniques – which can predict threats as
well as identify anomalies according to the features. Machine
learning is applicable in most of the common tasks including
classification, regression, and prediction; and thus appears as
FIGURE 15. Reported large scale distributed denial of service (DDoS)
attacks in different countries in 2007, Brazil was the most attacked one, a promising solution to cyber-vulnerabilities in this age of
while the United Kingdom and Japan stayed relatively safe. big data and lacking cyber-defense. Security at the smart
meters [218] is a good way to start on cyber protection. Also,
the general approach of most organization is to enforce
security at the smart meter, which is the tactical end point of
their responsibility area. But such approaches fail to realize
that the meter is not the only vulnerable area in the
infrastructure. Thus, it is imperative for organizations to
create a framework for assessing types of risks, and start this
evaluation from the very top: security concerns associated
with the strategy of their organization. A layered approach is
needed for securing the smart grid, and the direction of
strategy along with technical execution leads the way to such
security layers. The driving forces and requirements of
FIGURE 16. Costs of cyber-crimes in average around the world. The
United States faced a huge $17.6 million cost caused by such crimes, business process of any organization defines the strategic
more than double of what faced by the second most hurt country – direction; while technical execution embodies data privacy,
Japan.
security, data integrity, network security, physical security,
encryption, meter security, and associated operational
procedures. In a layered security framework, the data use and
security requirements are influenced by each layer according
to its responsibility and accountability [219]. This layered
security framework is demonstrated in fig. 18, where the
security considerations of each layer are indicated.

Author Name: Preparation of Papers for IEEE Access (February 2017)

characteristic uncertainty. This is the fundamental approach

for quantum computer-based cryptography. External
intervention or eavesdropping behavior will alter the state of
the qubits. It will ultimately disrupt the integrity of the
transferred information. Both private-key cryptography and
public-key cryptography requires key distribution. In private-
key cryptography, both ends of data transmission have a
shared key to encrypt and decrypt data. However, in public-
key cryptography, the key distribution is always done by a
public-key sever. A method using cryptography has been
proposed to approach the cyber security issue from modern
FIGURE 18. Layered security framework for smart grids. This quantum computing in [228].
comprehensive approach considers security at each stage of the
infrastructure, rather than only smart meter placed at consumer
location. Strategic direction and technical execution forms the two
major contributors in defining the framework responsibilities [219].
Since an IoT-centric cyber-physical system such as smart
grid provides a large volume of data, proper protection and
management of this data in an SG is very critical. From the
generation end to the distribution end, all kinds of data are
protected with various methods. Previously, a number of
work have been carried out on this purpose. Yuan et. al.
[220] developed a method for determining load distribution
attack behavior. Requirements and standards of cyber
security requirements have also been discussed previously in
references [221, 222]. Certain standards are also enacted by
various standardization organizations for cyber security
which cover diverse areas such as management of
information security, software size and quality, best
practices, cyber security outcomes, secure integrated
software and hardware testing, and industrial automation and
control systems. These are presented in table XIII. Data
aggregation in the AMI system [223] is a target for attacks as
well, and [224] presented a decentralized way of conducting
that task efficiently while maintaining data privacy. Cloud
computing is another important aspect of SG. This even
produced the term ‘cloud grid’ in China, which integrates the
nation’s power system with big data analysis facilities, IoT,
information and communication technologies, and of course,
cloud computing [225]. Security concerns regarding cloud
based infrastructure have been addressed previously in
references [226, 227]. Cryptography can provide protection
of data from security breach. Cryptography is a method
where data is stored in an encrypted manner. There are
several algorithms in use for carrying out this task of
encryption. They can be classified into various categories
according to their working principles. Table XIV presents
these security algorithms classified according to their
techniques and security-key lengths. Use of quantum
computer for encryption purpose can be very useful for data
security [228]. In a quantum computer, data is stored as
“qubits” rather than “bits”. From the Heisenberg’s
uncertainty principle, the values of momentum and position
of a physical system can be determined only with some

TABLE XIII
CURRENT CYBER-SECURITY STANDARDS AND STANDARDIZATION ORGANIZATIONS.
Standards Full Form Description

ISO/IEC ISO/IEC 27001:2013 – Information technology A standard for information security management system (ISMS).
27001 [229] – Security techniques – Information security
management systems – Requirements

CISQ [230] Consortium for IT Software Quality Standards to automate measurement of structural quality of software, and their size.

ISF [231] Information Security Forum Issued an inclusive list of information security best-practices named ‘Standard of Good
Practice (SoGP)’, updated every two years, except 2013-2014, the latest version came
out in 2016 [232].
NERC [233] North American Electric Reliability Corporation Aimed at identifying the source(s) utilized

NIST CSF National Institute of Standards and Technology Presents a high level taxonomy for outcomes of cybersecurity, along with a
[234] Cybersecurity Framework methodology for assessing and managing those outcomes.
ISO/IEC International Organization for Develops the ‘Common Criteria’ [236], and allows integration and secure testing of
15408 [235] Standardization/International Electrotechnical many different software and hardware products.
Commission 15408
RFC 2196 Request for Comments Developes security procedurs and policies for internet-connected information systems.
[237] Provides a broad and general overview of security of information which includes
security policies, incident response, and network security.
ISA/IEC International Society of Series of technical reports, standards, and related information to delineate procedures to
62443 [238] implement industrial automation and control systems (IACS) that are secure
Automation/International Electrotechnical
electronically. Applicable to system integrators, end-users, control systems
Commission 62443
manufacturers, and security practitioners responsible for designing, manufacturing,
managing, or implementing industrial automation and control systems.

TABLE XIV
ENCRYPTION ALGORITHMS FOR CYBER-SECURITY.
Category Name Key lengths for use between Key lengths for use now and beyond Reference
2011-2029 2030
(per SP 800-57 and SP 800-131) (per SP 800-57 and SP 800-131)
Symmetric Key Advanced Encryption Standard (AES) AES-128, AES-192, and AES-256 AES-128, AES-192, and AES-256 with [239]
with ECB, CBC, OFB, CFB-1, ECB, CBC, OFB, CFB-1, CFB-8, CFB-
CFB-8, CFB-128, CTR, or XTS 128, CTR, or XTS mode.
mode.

Triple-Data Encryption Algorithm 3-key TDES with TECB, TCBC, N/A – cannot use TDES beyond 2030 [240]
(TDEA) TCFB, TOFB, or CTR mode.

Asymetric Key Digital Signature DSA with (L=2048, N=224) or DSA with (L=3072, N=256) **, RSA [241]
Standard (DSS): (L=2048, N=256), RSA with with (|n|=3072) **, ECDSA2 with curves
Digital Signature Algorithm (DSA), (|n|=2048), ECDSA2 with curves P-256, P-384, P-521, K-283, K-409, K-
Rivest–Shamir–Adleman (RSA), P-224, K-233, or B-233 571, B-283, B-409, B-571
Elliptic Curve Digital Signature
Algorithm (ECDSA)
Secure Hash Secure Hash Algorithm (SHA) SHA-224 is approved for all SHA-256, SHA-384, and SHA-512 are [242]
Standard applications. Approved for all applications.

Message Cipher-based Message Authentication CMAC with 3-key TDES CMAC with AES-128, AES-192, or [243]
Authentication Code (CMAC) AES-256

Cipher block chaining - message All algorithms/key sizes listed in CCM with AES-128, AES-192, or AES- [244]
authentication code (CCM) the next column are approved. 256

Galois/Counter Mode (GCM)/ Galois All algorithms/key sizes listed in GCM with AES-128, AES-192, or AES- [245]
Message Authentication Code the next column are approved. 256
(GMAC)
Hash-based Message Authentication HMAC with SHA-1, SHA-224, HMAC with SHA-1, SHA-224, SHA- [246]
Code (HMAC) SHA-256, SHA-384, or SHA-512 256, SHA-384, or SHA-512 with Key
with 112 ≤ Key Length<128 bits Length ≥ 128 bits

251-253]. Ahmed et al. proposed a machine learning

For security measure in cloud computing based smart grid, approach based on Euclidean distance to detect FDIAs
a framework has been proposed by Baek et. al. in [247]. A [197]. They have also investigated on feature selection
flexible, scalable, and secure information management schemes with less complexity with improved accuracy that
framework with cloud computing topology has been employed genetic algorithm for bad data detection [201].
developed. The framework has three hierarchical levels: top, Ozay et al. used supervised learning to classify
regional and end-user levels. A brief description of all the measurement data as secure or compromised, and thus
levels of this hierarchy is depicted in table XV. For a secure detected FDIA in [254]. Their method was capable of
communication link between two different levels, identity of identifying attacks that are unobservable, and predict
the higher level can be used for the lower level to develop an attacks using observation sets. False data and stealth attack
detection in wide area measurement in smart grid
encrypted network [247]. The cloud computing centers have
monitoring system was demonstrated in [199]. Xin et al.
four major components: infrastructure-as-a-service (IaaS),
presented a detailed study on machine learning and deep
software-as-a-service (SaaS), platform-as-a-service (PaaS),
learning methods for intrusion detection in [255], where the
and data-as-a-service (DaaS). IaaS provides demand definitions of these areas are provided with descriptions of
response for all applications and services in the system. SaaS methods falling into each category. Wei et al. discussed on
provides services to the users; one such service can be detection of electricity theft in [202], and provided an
optimization of power. PaaS develops tools and libraries for overview on the works done in that sector. Machine
cloud computing applications. DaaS can be used for learning techniques such as principal component analysis
statistical purpose. There are four main clusters in the (PCA) [256], as well as game theory approaches such as the
proposed framework: information storage, general user Stackelberg game [257] can be applied in for detecting
services, control and management services, and electricity energy theft. Different types of software attacks and their
distribution services [56]. counter measures [44] are depicted in table XVI.
For the detection part of cyber-security, data TABLE XV
analysis, often employing machine learning is an obvious HIERARCHY OF SMART GRID SECURITY FRAMEWORK WITH CLOUD
COMPUTING TOPOLOGY.
choice for countering cyber-attacks in the data-driven Levels Responsibility
architecture of smart grid, and thus it has been heavily Top Accumulates data from regional cloud computing centers.
investigated in contemporary literature. Traditional Regional Manages intelligent devices.
signature based manual methods are almost useless in the End-users Provides data for regional level.
current complex systems, and machine learning has non-
linear analysis capabilities to detect false data injection in
complex systems [200]. Efficient threat detection is
particularly crucial for smart grids because of their
sensitivity to delays: the system gets exposed to higher
risks as the threat remains undetected for longer periods of
time [197]. Buczak et al. conducted a detailed study on data
mining and machine learning methods used for intrusion
detection [248]. They identified three major types of
intrusion detection systems: signature-based (detects
attacks from their signatures), anomaly-based (detects
attacks by deviation from normal system behavior), and
hybrid (combination of misused-based and anomaly-based
methods). Further information on anomaly detection can be
found in [249], while additional comparison of machine
learning techniques for intrusion detection is available in
[250]. Security threats to machine learning techniques
specifically can be found in [187].
For detecting false data insertion attacks (FDIA), general
machine learning techniques artificial neural network
(ANN) and support vector machines (SVM) were used
previously, while implementation of other techniques in
such detection were also conducted. Wang et al. employed
margin setting algorithm (MSA) which claimed better
results than the two methods mentioned before [200]. Other
notable techniques used for this cause are Bayesian
framework, particle swarm optimization (PSO), adaboost,
random forests, and common path mining method [200,

TABLE XVI
DIFFERENT SOFTWARE ATTACKS AND THEIR COUNTER MEASURES [44, 202].
Attack Type Description Counter Measure
Spoofing Unauthorized user can have access to a user's information; attacker may delete, change Introducing strong authentication
or control the information. mechanism
Password encryption
Using secure communication
protocol
Tampering Attackers modify user policies and Strong authorization
device parameters; possibility of harming people physically Digital signal
Secure communication link
Stackelberg game
Information disclosure User privacy can be manipulated. Strong authorization
Password encryption
Introducing private-enhanced
protocols
Denial of service (DoS) & Stopping all communications between stake holders, denial of access to EMS may be Use of home gateway to filter
Distributed denial of possible. address with the help of firewall
service (DDoS) Using trolling technique
Strong authorization
Honeypot models
Elevation of privileges The EMS includes third party plug-ins; which allow a sandboxed space in the EMS’s Assigning minimum role for users
functionality. When a malicious plug-in finds a backdoor, it could compromise EMS’s Accurate working of entire system
assets.
False data injection attack Injecting false data in the system. Using secure encryption techniques
(FDIA)
Unauthorized Access Gaining access to a program, service, server, website, or other system by means of Installing both spyware and virus
others’ accounts or some other method. protection programs
Protect sensitive data including
passwords and credit card
information
Traffic Analysis A special type of inference attack technique, monitoring communication patterns Dummy traffic approach to prevent
among entities of a system. traffic analysis attack employing
Friend in the Middle (FiM)
Eavesdropping Unauthorized interception of a private communication, for example instant message, Encryption
videoconference, phone call, or fax transmission, in real-time .
Masquerading Attacker pretending to be an authorized user for gaining access to a system. Enhanced key management
systems
Reply attacks Fraudulently or maliciously delaying or repeating a valid data transmission. Also Timestamping
known as playback attack.
Message modification and Modifying data on a target machine or direct a message to an alternate destination by Use of web application firewall
injection altering packet header addresses. Regular software patches
Suppressing error messages
Man-in-the-middle Secret relaying and possible alternation of communication between two parties who Secure or multipurpose internet
(MITM) attacks believe they are directly communicating with each other by an attacker in the middle. mail extensions
Authentication certificates
Flooding A form of denial-of-service attack, executed by sending a succession of SYN Filtering
(synchronize) requests to a target's system aimed at consuming enough server Increasing backlog
resources in order to make the system unresponsive to legitimate traffic. TCP half-open
Radio frequency(RF) Severe Denial-of-service attacks aimed at wireless medium. The attacker targets data Anti-jamming technologies
jamming packets of high importance by emitting radio frequency signals and do not follow
underlying network architecture.
Vulnerability attacks Vulnerability is a weakness that allows an attacker to reduce the information assurance Host-based intrusion detection
of a system. Vulnerabilities appear when three conditions meet: presence of system system
susceptibility or flaw, access of an attacker to that susceptibility, and the attacker’s Use of web proxy
capability to exploit this susceptibility. Use of accounts without
administrative privileges

government initiatives, and their approach to

VII. OUTCOMES
implementation of smart grid can also be studied.
The findings of this paper can be summarized as following:
 IoT devices can be worked on to make them more
 The electricity grid is currently going through the first
compact, cheap, energy efficient, and robust.
major change from its inception almost two centuries
Advanced communication protocols can also be
earlier. This next-generation grid system is combining
investigated to improve throughput and security.
power system, information technology,
Monitoring schemes of power generation facility,
communication and control systems to create a robust
pumps, and turbines can be further developed.
and adaptive infrastructure better suited to
accommodate new and emerging technologies. This  Better forecasting techniques for demand and
new grid is called the smart grid. generation, especially renewable energy generation is
essential for proper operation of renewable-energy
 Connectivity and exchange of information lies at the
based smart grids.
core of smart grid functionality, which made
connected devices a corner-stone for this technology.  Machine learning algorithms can be developed to
These devices are called the “internet of things (IoT)”, meet power quality standards in a smart grid using the
and enable the grid components to exchange data to available data. Machine learning algorithms can also
maintain an up-to-date system status and receive be applied in wind-solar hybrid system to further
commands to act as grid conditions change. IoT utilize our available resources.
devices are increasing significantly in number each  More research is required to develop viable solutions
year, and are bringing unique opportunities and for other security concerns such as physical threats,
challenges with their wider implementation. network attacks and encryption attacks.
 IoT devices generate a huge amount of data, which Communication systems also need to be more
cannot be handled through conventional analysis efficient, with more protective measure.
techniques. This massive data is termed as “big data”,
VIII. CONCLUSION
and it motivated the move towards new data analysis
techniques. Big data generated from IoT devices are The electricity grid is transitioning towards an IoT-based,
also exposed to security threats, and that have connected smart grid, and with the benefits of such a system,
attracted a lot of attention as well. concerns are also emerging that were unprecedented until
 Machine learning is a useful way to sift through big now. The big data generated in the smart grid is requiring
data, and extract useful information that can novel analysis techniques such as machine learning methods
extensively aid in demand and generation pattern for proper handling and data extraction. The connected
recognition, generation forecasting, control etc. A devices, and the data they generate are also bringing forth the
number of methods have already been presented in dire necessities of proper protection, as they are being
existing literature, and more novel techniques are targeted to attacks of varying magnitudes which highlighted
being worked on for enhanced performance in specific the lack of proper counter-measures in place. In an attempt to
use cases. present an overall picture of these issues, this paper had
 Every sector of the smart grid – generation, presented a brief timeline of the grid’s journey to the smart
transmission, and distribution – are in significant risk grid, and how internet of things (IoT) had become a part and
of cyber-attacks, and many such attacks have already parcel of the electricity grid. Challenges associated with IoT-
been carried out. Security of data is thus a major generated big data, namely their analysis and protection, as
concern in smart grid, and significant amount of work well as other security concerns in the smart grid had also
has already been conducted on detection of cyber- been discussed. The outcomes of this study had been
security threats and protection mechanisms to counter presented finally with future research directions outlined
them. Many of these counter-measures have used briefly to aid researchers in this field.
machine learning techniques, as conventional methods
are often useless in the new data-centric, non-linear REFERENCES
system. [1] L. M. Camarinha-Matos, "Collaborative smart grids–A
survey on trends," Renewable and Sustainable Energy
Based on this study, the following can be stated for future Reviews, vol. 65, pp. 283-294, 2016.
works in this field: [2] R. Bayindir, E. Hossain, E. Kabalci, and R. Perez, "A
 The viability of the current grid infrastructure need to comprehensive study on microgrid technology,"
be validated through approaches such as mathematical
International Journal of Renewable Energy Research, vol. 4,
modelling to find out the optimum timeframe and
pp. 1094-1107, 2014.
technological approach to move towards the smart
[3] E. Hossain, R. Perez, S. Padmanaban, and P. Siano,
grid architecture.
"Investigation on the Development of a Sliding Mode
 The challenges in transitioning to the renewable
Controller for Constant Power Loads in Microgrids,"
energy-centric smart grid, and their feasible solutions
Energies, vol. 10, p. 1086, 2017.
need to be investigated. Possible business models,

VOLUME XX, 2017 9

[4] E. Hossain, E. Kabalci, R. Bayindir, and R. Perez, [20] S. Aman, Y. Simmhan, and V. K. Prasanna, "Holistic
"Microgrid testbeds around the world: State of art," Energy measures for evaluating prediction models in smart grids,"
Conversion and Management, vol. 86, pp. 132-153, 2014. IEEE Transactions on Knowledge and Data Engineering,
[5] W.-Y. Chiu, H. Sun, and H. V. Poor, "A multiobjective vol. 27, pp. 475-488, 2015.
approach to multimicrogrid system design," IEEE [21] S. Aman, M. Frincu, C. Chelmis, M. Noor, Y. Simmhan, and
Transactions on Smart Grid, vol. 6, pp. 2263-2272, 2015. V. K. Prasanna, "Prediction models for dynamic demand
[6] J. Rodríguez-Molina, M. Martínez-Núñez, J.-F. Martínez, response: Requirements, challenges, and insights," in Smart
and W. Pérez-Aguiar, "Business models in the smart grid: Grid Communications (SmartGridComm), 2015 IEEE
Challenges, opportunities and proposals for prosumer International Conference on, 2015, pp. 338-343.
profitability," Energies, vol. 7, pp. 6142-6171, 2014. [22] M. Manfren, "Multi-commodity network flow models for
[7] J. Yuan, J. Shen, L. Pan, C. Zhao, and J. Kang, "Smart grids dynamic energy management–Mathematical formulation,"
in China," Renewable and Sustainable Energy Reviews, vol. Energy Procedia, vol. 14, pp. 1380-1385, 2012.
37, pp. 896-906, 2014. [23] P. D. Diamantoulakis, V. M. Kapinas, and G. K.
[8] B. L. R. Stojkoska and K. V. Trivodaliev, "A review of Karagiannidis, "Big data analytics for dynamic energy
Internet of Things for smart home: Challenges and management in smart grids," Big Data Research, vol. 2, pp.
solutions," Journal of Cleaner Production, vol. 140, pp. 94-101, 2015.
1454-1464, 2017. [24] J. Yuan, S. Sun, J. Shen, Y. Xu, and C. Zhao, "Wind power
[9] D. S. Markovic, I. Branovic, and R. Popovic, "Smart Grid supply chain in China," Renewable and Sustainable Energy
and nanotechnologies: a solution for clean and sustainable Reviews, vol. 39, pp. 356-369, 2014.
energy," Energy Emission Control Technol, vol. 3, pp. 1-13, [25] D. A. Powner, Electricity Grid Modernization: Progress
2015. Being Made on Cybersecurity Guidelines, But Key
[10] W. Yu, G. Wen, X. Yu, Z. Wu, and J. Lü, "Bridging the gap Challenges Remain to be Addressed: DIANE Publishing,
between complex networks and smart grids," Journal of 2011.
Control and Decision, vol. 1, pp. 102-114, 2014. [26] S. Simitis, "From the market to the polis: The EU directive
[11] S. Abdollahy, A. Mammoli, F. Cheng, A. Ellis, and J. on the protection of personal data," Iowa L. Rev., vol. 80, p.
Johnson, "Distributed compensation of a large intermittent 445, 1994.
energy resource in a distribution feeder," in Innovative Smart [27] J. Hu and A. V. Vasilakos, "Energy big data analytics and
Grid Technologies (ISGT), 2013 IEEE PES, 2013, pp. 1-6. security: challenges and opportunities," IEEE Transactions
[12] J. P. Lopes, C. Moreira, and A. Madureira, "Defining control on Smart Grid, vol. 7, pp. 2423-2436, 2016.
strategies for analysing microgrids islanded operation," in [28] P. Leitao, S. Karnouskos, L. Ribeiro, J. Lee, T. Strasser, and
Power Tech, 2005 IEEE Russia, 2005, pp. 1-7. A. W. Colombo, "Smart agents in industrial cyber–physical
[13] H. Gharavi and R. Ghafurian, "Smart Grid: The Electric systems," Proceedings of the IEEE, vol. 104, pp. 1086-1101,
Energy System of the Future [Scanning the Issue]," 2016.
Proceedings of the IEEE, vol. 99, pp. 917-921, 2011. [29] M. Safiuddin, HISTORY OF ELECTRIC GRID, 2013.
[14] A. Amato, R. Aversa, B. Di Martino, and S. Venticinque, "A [30] E. Hossain, R. Perez, A. Nasiri, and S. Padmanaban, "A
cyber physical system of smart micro-grids," in Network- Comprehensive Review on Constant Power Loads
Based Information Systems (NBiS), 2016 19th International Compensation Techniques," IEEE Access, vol. 6, pp. 33285-
Conference on, 2016, pp. 165-172. 33305, 2018.
[15] M. Kezunovic, "Data Analytics: Creating Information and [31] J. Gubbi, R. Buyya, S. Marusic, and M. Palaniswami,
Knowledge [Guest Editorial]," IEEE Power and Energy "Internet of Things (IoT): A vision, architectural elements,
Magazine, vol. 10, pp. 14-23, 2012. and future directions," Future generation computer systems,
[16] D. Alahakoon and X. Yu, "Smart electricity meter data vol. 29, pp. 1645-1660, 2013.
intelligence for future energy systems: A survey," IEEE [32] A. Nordrum, "Popular internet of things forecast of 50
Transactions on Industrial Informatics, vol. 12, pp. 425-436, billion devices by 2020 is outdated," IEEE Spectrum, vol.
2016. 18, 2016.
[17] J. Shishido and E. U. Solutions, "Smart meter data quality [33] L. Liu, Y. Liu, L. Wang, A. Zomaya, and S. Hu,
insights," in ACEEE Summer Study on Energy Efficiency in "Economical and balanced energy usage in the smart home
Buildings, 2012. infrastructure: A tutorial and new results," IEEE
[18] A. McKane, I. Rhyne, A. Lekov, L. Thompson, and M. A. Transactions on Emerging Topics in Computing, vol. 3, pp.
Piette, "Automated demand response: The missing link in 556-570, 2015.
the electricity value chain," August 2009. [34] C. Gamarra, J. M. Guerrero, and E. Montero, "A knowledge
[19] X. Yu, C. Cecati, T. Dillon, and M. G. Simoes, "The new discovery in databases approach for industrial microgrid
frontier of smart grids," IEEE Industrial Electronics planning," Renewable and Sustainable Energy Reviews, vol.
Magazine, vol. 5, pp. 49-63, 2011. 60, pp. 615-630, 2016.

VOLUME XX, 2017 9

[35] M. Khan, B. N. Silva, and K. Han, "Internet of things based power distribution systems," Computers & Electrical
energy aware smart home control system," IEEE Access, vol. Engineering, vol. 52, pp. 183-198, 2016.
4, pp. 7556-7566, 2016. [50] A. Al-Ali, I. A. Zualkernan, M. Rashid, R. Gupta, and M.
[36] Y. Saleem, N. Crespi, M. H. Rehmani, and R. Copeland, Alikarar, "A smart home energy management system using
"Internet of things-aided Smart Grid: technologies, IoT and big data analytics approach," IEEE Transactions on
architectures, applications, prototypes, and future research Consumer Electronics, vol. 63, pp. 426-434, 2017.
directions," arXiv preprint arXiv:1704.08977, 2017. [51] F. Un-Noor, S. Padmanaban, L. Mihet-Popa, M. Mollah, and
[37] J. Pan, R. Jain, S. Paul, T. Vu, A. Saifullah, and M. Sha, "An E. Hossain, "A Comprehensive Study of Key Electric
internet of things framework for smart energy in buildings: Vehicle (EV) Components, Technologies, Challenges,
designs, prototype, and experiments," IEEE Internet of Impacts, and Future Direction of Development," Energies,
Things Journal, vol. 2, pp. 527-537, 2015. vol. 10, p. 1217, 2017.
[38] M. Ligade. (2016, April 9). Architecture for IOT [52] Z. Zhu, S. Lambotharan, W. H. Chin, and Z. Fan, "Overview
applications. Available: of demand management in smart grid and enabling wireless
https://medium.com/@maheshwar.ligade/architecture-for- communication technologies," IEEE Wireless
iot-applications-d50ece031d38 Communications, vol. 19, 2012.
[39] J. Bauernberger. (2016, April 9). Securing the Internet of [53] (2015, March 1). RTI Presents "How to Architect Microgrids
Things. Available: https://blog.valbonne- for the Industrial Internet of Things" Complimentary
consulting.com/2016/04/22/securing-the-internet-of-things/ Webinar. Available: https://www.rti.com/news/how-to-
[40] S. Thakurdesai, "Smarter Metering - Now & in Future," architect-microgrids-webinar
Texas Instruments. [54] M. Rana and L. Li, "Microgrid state estimation and control
[41] A. Basit, G. A. S. Sidhu, A. Mahmood, and F. Gao, for smart grid and Internet of Things communication
"Efficient and autonomous energy management techniques network," Electronics Letters, vol. 51, pp. 149-151, 2015.
for the future smart homes," IEEE Transactions on Smart [55] M. M. Rana and L. Li, "An overview of distributed
Grid, vol. 8, pp. 917-926, 2017. microgrid state estimation and control for smart grids,"
[42] W. Shu-wen, "Research on the key technologies of IOT Sensors, vol. 15, pp. 4302-4325, 2015.
applied on Smart Grid," in Electronics, Communications and [56] S. Kinney. (2016, March 1). The smart grid and its key role
Control (ICECC), 2011 International Conference on, 2011, in the Industrial IoT. Available:
pp. 2809-2812. https://enterpriseiotinsights.com/20160418/channels/use-
[43] Y. Wang, W. Lin, T. Zhang, and Y. Ma, "Research on cases/smart-grids-key-role-industrial-iot
application and security protection of internet of things in [57] R. Bikmetov, M. Y. A. Raja, and T. U. Sane, "Infrastructure
smart grid," 2012. and applications of Internet of Things in smart grids: A
[44] Q. Ou, Y. Zhen, X. Li, Y. Zhang, and L. Zeng, "Application survey," in Power Symposium (NAPS), 2017 North
of internet of things in smart grid power transmission," in American, 2017, pp. 1-6.
Mobile, Ubiquitous, and Intelligent Computing (MUSIC), [58] H. Mortaji, S. H. Ow, M. Moghavvemi, and H. A. F.
2012 Third FTRA International Conference on, 2012, pp. Almurib, "Load Shedding and Smart-Direct Load Control
96-100. Using Internet of Things in Smart Grid Demand Response
[45] M. M. Rogers, G. Xu, C. J. Miller, S. P. McElmurry, W. Shi, Management," IEEE Transactions on Industry Applications,
Y. Wang, et al., "HERO: A smart-phone application for vol. 53, pp. 5155-5163, 2017.
location based emissions estimates," Sustainable Computing: [59] D. Minoli, K. Sohraby, and B. Occhiogrosso, "IoT
Informatics and Systems, vol. 8, pp. 3-7, 2015. Considerations, Requirements, and Architectures for Smart
[46] B. Martinez, M. Monton, I. Vilajosana, and J. D. Prades, Buildings—Energy Optimization and Next-Generation
"The power of models: Modeling power consumption for Building Management Systems," IEEE Internet of Things
IoT devices," IEEE Sensors Journal, vol. 15, pp. 5777-5789, Journal, vol. 4, pp. 269-283, 2017.
2015. [60] A. F. Taha, J. Qi, J. Wang, and J. H. Panchal, "Risk
[47] J. Lin, C. Wan, Y. Song, R. Huang, X. Chen, W. Guo, et al., Mitigation for Dynamic State Estimation Against Cyber
"Situation awareness of active distribution network: Attacks and Unknown Inputs," IEEE Transactions on Smart
roadmap, technologies, and bottlenecks," CSEE Journal of Grid, vol. 9, pp. 886-899, 2018.
Power and Energy Systems, vol. 2, pp. 35-42, 2016. [61] Y. Seyedi and H. Karimi, "Coordinated protection and
[48] V. Madani, R. Das, F. Aminifar, J. McDonald, S. Venkata, control based on synchrophasor data processing in smart
D. Novosel, et al., "Distribution automation strategies distribution networks," IEEE Transactions on Power
challenges and opportunities in a changing landscape," IEEE Systems, vol. 33, pp. 634-645, 2018.
Transactions on Smart Grid, vol. 6, pp. 2157-2165, 2015. [62] Y. Zhang, Y. Xu, and Z. Y. Dong, "Robust classification
[49] S. A. Amin, A. Ali-Eldin, and H. A. Ali, "A context-aware model for PMU-based on-line power system DSA with
dispatcher for the Internet of Things: The case of electric missing data," IET Generation, Transmission &
Distribution, vol. 11, pp. 4484-4491, 2017.

VOLUME XX, 2017 9

[63] B. K. Tannahill and M. Jamshidi, "System of Systems and [78] L. Collins and J. Ward, "Real and reactive power control of
Big Data analytics–Bridging the gap," Computers & distributed PV inverters for overvoltage prevention and
Electrical Engineering, vol. 40, pp. 2-15, 2014. increased renewable generation hosting capacity,"
[64] D. Tayal, "Disruptive forces on the electricity industry: A Renewable Energy, vol. 81, pp. 464-471, 2015.
changing landscape for utilities," The Electricity Journal, [79] J. Soares, N. Borges, M. A. F. Ghazvini, Z. Vale, and P. de
vol. 29, pp. 13-17, 2016. Moura Oliveira, "Scenario generation for electric vehicles'
[65] K. Wang, Y. Wang, X. Hu, Y. Sun, D.-J. Deng, A. Vinel, et uncertain behavior in a smart city environment," Energy,
al., "Wireless big data computing in smart grid," IEEE vol. 111, pp. 664-675, 2016.
Wireless Communications, vol. 24, pp. 58-64, 2017. [80] E.-K. Lee, W. Shi, R. Gadh, and W. Kim, "Design and
[66] N. Zhang, Y. Yan, S. Xu, and W. Su, "A distributed data Implementation of a Microgrid Energy Management
storage and processing framework for next-generation System," Sustainability, vol. 8, p. 1143, 2016.
residential distribution systems," Electric Power Systems [81] N.-C. Yang and W.-C. Tseng, "Adaptive three-phase power-
Research, vol. 116, pp. 174-181, 2014. flow solutions for smart grids with plug-in hybrid electric
[67] A. Paul, A. Ahmad, M. M. Rathore, and S. Jabbar, vehicles," International Journal of Electrical Power &
"Smartbuddy: defining human behaviors using big data Energy Systems, vol. 64, pp. 1166-1175, 2015.
analytics in social internet of things," IEEE Wireless [82] K. Zhou, S. Yang, and Z. Shao, "Energy internet: the
Communications, vol. 23, pp. 68-74, 2016. business perspective," Applied Energy, vol. 178, pp. 212-
[68] N. Uribe-Pérez, L. Hernández, D. de la Vega, and I. Angulo, 222, 2016.
"State of the art and trends review of smart metering in [83] M. Peters, W. Ketter, M. Saar-Tsechansky, and J. Collins,
electricity grids," Applied Sciences, vol. 6, p. 68, 2016. "A reinforcement learning approach to autonomous decision-
[69] Z. Guan, J. Li, L. Wu, Y. Zhang, J. Wu, and X. Du, making in smart electricity markets," Machine learning, vol.
"Achieving Efficient and Secure Data Acquisition for Cloud- 92, pp. 5-39, 2013.
Supported Internet of Things in Smart Grid," IEEE Internet [84] K. Zhou and S. Yang, "Understanding household energy
of Things Journal, vol. 4, pp. 1934-1944, 2017. consumption behavior: The contribution of energy big data
[70] B. K. Tannahill and M. Jamshidi, "Big data analytic analytics," Renewable and Sustainable Energy Reviews, vol.
paradigms-from pca to deep learning," in Proceedings of the 56, pp. 810-819, 2016.
2014 The Association for the Advancement of Artificial [85] K. Zhou and S. Yang, "Demand side management in China:
Intelligence Symposium (AAAI 2014), 2014, pp. 84-90. The context of China’s power industry reform," Renewable
[71] A. Emrouznejad and A. Emrouznejad, Big Data and Sustainable Energy Reviews, vol. 47, pp. 954-965, 2015.
Optimization: Recent Developments and Challenges vol. 18: [86] K. Zhou, C. Fu, and S. Yang, "Big data driven smart energy
Springer, 2016. management: From big data to big insights," Renewable and
[72] W. Pedrycz and S.-M. Chen, Data science and big data: an Sustainable Energy Reviews, vol. 56, pp. 215-225, 2016.
environment of computational intelligence vol. 24: Springer, [87] (April 10). How can we differentiate Big Data Analytics
2017. from Statistical Predictive Modeling Techniques? Available:
[73] L. Rodríguez-Mazahua, C.-A. Rodríguez-Enríquez, J. L. http://www.zarantech.com/blog/can-differentiate-big-data-
Sánchez-Cervantes, J. Cervantes, J. L. García-Alcaraz, and analytics-statistical-predictive-modelling-techniques/
G. Alor-Hernández, "A general perspective of Big Data: [88] S. Mujawar and A. Joshi, "Data Analytics Types Tools and
applications, tools, challenges and trends," The Journal of their Comparison," IIJARCE, vol. 4, pp. 488-491, 2015.
Supercomputing, vol. 72, pp. 3073-3113, 2016. [89] P. Hassani. (2016, April 10). An Insight into 26 Big Data
[74] C. Miller, M. Martin, D. Pinney, and G. Walker, "Achieving Analytic Techniques: Part 1. Available:
a resilient and agile grid," National Rural Electric https://blogs.systweak.com/2016/11/an-insight-into-26-big-
Cooperative Association (NRECA): Arlington, VA, USA, data-analytic-techniques-part-1/
2014. [90] X. He, Q. Ai, R. C. Qiu, W. Huang, L. Piao, and H. Liu, "A
[75] J. Giri, "Proactive management of the future grid," IEEE big data architecture design for smart grids based on random
Power and Energy Technology Systems Journal, vol. 2, pp. matrix theory," IEEE transactions on smart Grid, vol. 8, pp.
43-52, 2015. 674-686, 2017.
[76] S. Aman, Y. Simmhan, and V. K. Prasanna, "Energy [91] S. Ryu, J. Noh, and H. Kim, "Deep neural network based
management systems: state of the art and emerging trends," demand side short term load forecasting," Energies, vol. 10,
IEEE Communications Magazine, vol. 51, pp. 114-119, p. 3, 2016.
2013. [92] I. M. Coelho, V. N. Coelho, E. J. d. S. Luz, L. S. Ochi, F. G.
[77] M. T. Burr, M. J. Zimmer, G. Warner, B. Meloy, J. Bertrand, Guimarães, and E. Rios, "A GPU deep learning
W. Levesque, et al., "Emerging Models for Microgrid metaheuristic based model for time series forecasting,"
Finance: Driven by the need to deliver value to end users," Applied Energy, vol. 201, pp. 412-418, 2017.
IEEE Electrification Magazine, vol. 2, pp. 30-39, 2014. [93] R. Bessa, A. Trindade, C. S. Silva, and V. Miranda,
"Probabilistic solar power forecasting in smart grids using

VOLUME XX, 2017 9

distributed information," International Journal of Electrical [107] M. Ali, Z. Dong, X. Li, and P. Zhang, "RSA-Grid: A grid
Power & Energy Systems, vol. 72, pp. 16-23, 2015. computing based framework for power system reliability and
[94] (April 10). Top 10 Data Science, Machine Learning Methods security analysis," in Power Engineering Society General
Used, 2017. Available: Meeting, 2006. IEEE, 2006, p. 7 pp.
https://www.kdnuggets.com/2017/12/top-data-science- [108] X. Wang, Q. Liang, J. Mu, W. Wang, and B. Zhang,
machine-learning-methods.html "Physical layer security in wireless smart grid," Security and
[95] V. G. Agelidis, C. Pavlovski, M. Ciobotaru, T. Communication Networks, vol. 8, pp. 2431-2439, 2015.
Tschirschwitz, R. Corke, T. Lampard, et al., "Unlocking the [109] B. Dickson. (2016, March 1). Exploiting machine learning in
Smart Grid: An Australian industry-university collaborative cybersecurity. Available:
effort to address skill gaps," in Innovative Smart Grid https://techcrunch.com/2016/07/01/exploiting-machine-
Technologies Asia (ISGT), 2011 IEEE PES, 2011, pp. 1-8. learning-in-cybersecurity/
[96] A. Sanchez and W. Rivera, "Big Data Analysis and [110] M. Frincu, C. Chelmis, M. U. Noor, and V. Prasanna,
Visualization for the Smart Grid," in Big Data (BigData "Accurate and efficient selection of the best consumption
Congress), 2017 IEEE International Congress on, 2017, pp. prediction method in smart grids," in Big Data (Big Data),
414-418. 2014 IEEE International Conference on, 2014, pp. 721-729.
[97] M. He, S. Murugesan, and J. Zhang, "Multiple timescale [111] M. Esmalifalak, L. Liu, N. Nguyen, R. Zheng, and Z. Han,
dispatch and scheduling for stochastic reliability in smart "Detecting stealthy false data injection using machine
grids with wind generation integration," in INFOCOM, 2011 learning in smart grid," IEEE Systems Journal, 2014.
Proceedings IEEE, 2011, pp. 461-465. [112] W. Liu, B. Tang, J. Han, X. Lu, N. Hu, and Z. He, "The
[98] D. Niyato, E. Hossain, and A. Fallahi, "Sleep and wakeup structure healthy condition monitoring and fault diagnosis
strategies in solar-powered wireless sensor/mesh networks: methods in wind turbines: A review," Renewable and
Performance analysis and optimization," IEEE Transactions Sustainable Energy Reviews, vol. 44, pp. 466-472, 2015.
on Mobile Computing, vol. 6, pp. 221-236, 2007. [113] A. Kusiak and Z. Zhang, "Adaptive control of a wind turbine
[99] H. Jiang, X. Dai, D. W. Gao, J. J. Zhang, Y. Zhang, and E. with data mining and swarm intelligence," IEEE
Muljadi, "Spatial-temporal synchrophasor data Transactions on Sustainable Energy, vol. 2, pp. 28-36, 2011.
characterization and analytics in smart grid fault detection, [114] A. Mellit, S. A. Kalogirou, L. Hontoria, and S. Shaari,
identification, and impact causal analysis," IEEE "Artificial intelligence techniques for sizing photovoltaic
Transactions on Smart Grid, vol. 7, pp. 2525-2536, 2016. systems: A review," Renewable and Sustainable Energy
[100] D. Li and S. K. Jayaweera, "Machine-learning aided optimal Reviews, vol. 13, pp. 406-419, 2009.
customer decisions for an interactive smart grid," IEEE [115] M. Negnevitsky, P. Mandal, and A. K. Srivastava, "Machine
Systems Journal, vol. 9, pp. 1529-1540, 2015. learning applications for load, price and wind power
[101] Y. Tang, J. Yang, J. Yan, and H. He, "Intelligent load prediction in power systems," in Intelligent System
frequency controller using GrADP for island smart grid with Applications to Power Systems, 2009. ISAP'09. 15th
electric vehicles and renewable resources," Neurocomputing, International Conference on, 2009, pp. 1-6.
vol. 170, pp. 406-416, 2015. [116] L. Wenyi, W. Zhenfeng, H. Jiguang, and W. Guangfeng,
[102] X. Fang, D. Yang, and G. Xue, "Online strategizing "Wind turbine fault diagnosis method based on diagonal
distributed renewable energy resource access in islanded spectrum and clustering binary tree SVM," Renewable
microgrids," in Global Telecommunications Conference Energy, vol. 50, pp. 1-6, 2013.
(GLOBECOM 2011), 2011 IEEE, 2011, pp. 1-6. [117] Y. Xu, Z. Dong, K. Meng, R. Zhang, and K. Wong, "Real-
[103] D. O'Neill, M. Levorato, A. Goldsmith, and U. Mitra, time transient stability assessment model using extreme
"Residential demand response using reinforcement learning," learning machine," IET generation, transmission &
in Smart Grid Communications (SmartGridComm), 2010 distribution, vol. 5, pp. 314-322, 2011.
First IEEE International Conference on, 2010, pp. 409-414. [118] B. Wang, B. Fang, Y. Wang, H. Liu, and Y. Liu, "Power
[104] H. Wang, Q. Wang, P. Liu, and L. Sun, "Big data and system transient stability assessment based on big data and
intelligent agent based smart grid architecture," in Agents the core vector machine," IEEE Transactions on Smart Grid,
(ICA), 2017 IEEE International Conference on, 2017, pp. vol. 7, pp. 2561-2570, 2016.
106-107. [119] B. Schölkopf and A. J. Smola, Learning with kernels:
[105] T. Chen, Y. Zhang, X. Wang, and G. B. Giannakis, "Robust support vector machines, regularization, optimization, and
workload and energy management for sustainable data beyond: MIT press, 2002.
centers," IEEE Journal on Selected Areas in [120] P. Zhao, J. Xia, Y. Dai, and J. He, "Wind speed prediction
Communications, vol. 34, pp. 651-664, 2016. using support vector regression," in Industrial Electronics
[106] R. C. Green, L. Wang, and M. Alam, "Applications and and Applications (ICIEA), 2010 the 5th IEEE Conference
trends of high performance computing for electric power on, 2010, pp. 882-886.
systems: Focusing on smart grid," IEEE Transactions on
Smart Grid, vol. 4, pp. 922-931, 2013.

VOLUME XX, 2017 9

[121] J. Zeng and W. Qiao, "Short-term solar power prediction time energy management," in Big Data, 2013 IEEE
using a support vector machine," Renewable Energy, vol. 52, International Conference on, 2013, pp. 657-664.
pp. 118-127, 2013. [135] D. François, "Methodology and standards for data analysis
[122] H. A. Kazem, J. H. Yousif, and M. T. Chaichan, "Modeling with machine learning tools," 2008.
of daily solar energy system prediction using support vector [136] Formulated.by. (2018, April 10). AI and Machine Learning
machine for Oman," International Journal of Applied in Cyber Security. Available:
Engineering Research, vol. 11, pp. 10166-10172, 2016. https://towardsdatascience.com/ai-and-machine-learning-in-
[123] J.-L. Chen, H.-B. Liu, W. Wu, and D.-T. Xie, "Estimation of cyber-security-d6fbee480af0
monthly solar radiation from measured temperatures using [137] "Global Wind 2006 Report," ed. Brussels, Belgium: Global
support vector machines–a case study," Renewable Energy, Wind Energy Council.
vol. 36, pp. 413-420, 2011. [138] H. Lund and W. Kempton, "Integration of renewable energy
[124] Y. Y. Chia, L. H. Lee, N. Shafiabady, and D. Isa, "A load into the transport and electricity sectors through V2G,"
predictive energy management system for supercapacitor- Energy policy, vol. 36, pp. 3578-3587, 2008.
battery hybrid energy storage system in solar application [139] A. M. Foley, P. G. Leahy, A. Marvuglia, and E. J. McKeogh,
using the Support Vector Machine," Applied Energy, vol. "Current methods and advances in forecasting of wind power
137, pp. 588-602, 2015. generation," Renewable Energy, vol. 37, pp. 1-8, 2012.
[125] F. Liberati and A. Di Giorgio, "Economic Model Predictive [140] F. Douak, F. Melgani, and N. Benoudjit, "Kernel ridge
and Feedback Control of a Smart Grid Prosumer Node," regression with active learning for wind speed prediction,"
Energies, vol. 11, p. 48, 2017. Applied energy, vol. 103, pp. 328-340, 2013.
[126] F. Ucar, O. F. Alcin, B. Dandil, and F. Ata, "Power Quality [141] A. Clifton, L. Kilcher, J. Lundquist, and P. Fleming, "Using
Event Detection Using a Fast Extreme Learning Machine," machine learning to predict wind turbine power output,"
Energies, vol. 11, p. 145, 2018. Environmental research letters, vol. 8, p. 024009, 2013.
[127] L. Morales-Velazquez, R. de Jesus Romero-Troncoso, G. [142] Z. Zhou, F. Xiong, B. Huang, C. Xu, R. Jiao, B. Liao, et al.,
Herrera-Ruiz, D. Morinigo-Sotelo, and R. A. Osornio-Rios, "Game-Theoretical Energy Management for Energy Internet
"Smart sensor network for power quality monitoring in With Big Data-Based Renewable Power Forecasting," IEEE
electrical installations," Measurement, vol. 103, pp. 133-142, Access, vol. 5, pp. 5731-5746, 2017.
2017. [143] C. W. Potter and M. Negnevitsky, "Very short-term wind
[128] B. Li, S. Gangadhar, S. Cheng, and P. K. Verma, "Predicting forecasting for Tasmanian power generation," IEEE
user comfort level using machine learning for smart grid Transactions on Power Systems, vol. 21, pp. 965-972, 2006.
environments," in Innovative Smart Grid Technologies [144] T. G. Barbounis, J. B. Theocharis, M. C. Alexiadis, and P. S.
(ISGT), 2011 IEEE PES, 2011, pp. 1-6. Dokopoulos, "Long-term wind speed and power forecasting
[129] T. R, E. A. Jasmin, and T. P. I. Ahamed, "Residential Load using local recurrent neural network models," IEEE
Scheduling With Renewable Generation in the Smart Grid: Transactions on Energy Conversion, vol. 21, pp. 273-284,
A Reinforcement Learning Approach," IEEE Systems 2006.
Journal, pp. 1-12, 2018. [145] B. Zhu, M.-y. Chen, N. Wade, and L. Ran, "A prediction
[130] S. Alshareef, S. Talwar, and W. G. Morsi, "A new approach model for wind farm power generation based on fuzzy
based on wavelet design and machine learning for islanding modeling," Procedia Environmental Sciences, vol. 12, pp.
detection of distributed generation," IEEE Transactions on 122-129, 2012.
Smart Grid, vol. 5, pp. 1575-1583, 2014. [146] J. L. Torres, A. Garcia, M. De Blas, and A. De Francisco,
[131] H. Jiang, Y. Li, Y. Zhang, J. J. Zhang, D. W. Gao, E. "Forecast of hourly average wind speed with ARMA models
Muljadi, et al., "Big data-based approach to detect, locate, in Navarre (Spain)," Solar Energy, vol. 79, pp. 65-77, 2005.
and enhance the stability of an unplanned microgrid [147] M. C. Mabel and E. Fernandez, "Analysis of wind power
islanding," Journal of Energy Engineering, vol. 143, p. generation and prediction using ANN: A case study,"
04017045, 2017. Renewable energy, vol. 33, pp. 986-992, 2008.
[132] S. Jurado, À. Nebot, F. Mugica, and N. Avellana, "Hybrid [148] R. E. Abdel-Aal, M. A. Elhadidy, and S. Shaahid, "Modeling
methodologies for electricity load forecasting: Entropy- and forecasting the mean hourly wind speed time series
based feature selection with machine learning and soft using GMDH-based abductive networks," Renewable
computing techniques," Energy, vol. 86, pp. 276-291, 2015. Energy, vol. 34, pp. 1686-1699, 2009.
[133] C. Cecati, J. Kolbusz, P. Różycki, P. Siano, and B. M. [149] M. Yesilbudak, S. Sagiroglu, and I. Colak, "A new approach
Wilamowski, "A novel RBF training algorithm for short- to very short term wind speed prediction using k-nearest
term electric load forecasting and comparative studies," neighbor classification," energy conversion and
IEEE Transactions on industrial Electronics, vol. 62, pp. management, vol. 69, pp. 77-86, 2013.
6519-6529, 2015. [150] L. Li, Y.-q. LIU, Y.-p. YANG, H. Shuang, and Y.-m.
[134] N. Balac, T. Sipes, N. Wolter, K. Nunes, B. Sinkovits, and WANG, "A physical approach of the short-term wind power
H. Karimabadi, "Large scale predictive analytics for real-

VOLUME XX, 2017 9

prediction based on CFD pre-calculated flow fields," Journal [164] Y. Zhang, K. Liu, L. Qin, and X. An, "Deterministic and
of Hydrodynamics, Ser. B, vol. 25, pp. 56-61, 2013. probabilistic interval prediction for short-term wind power
[151] K. Mohammadi, S. Shamshirband, L. Yee, D. Petković, M. generation based on variational mode decomposition and
Zamani, and S. Ch, "Predicting the wind power density machine learning methods," Energy Conversion and
based upon extreme learning machine," Energy, vol. 86, pp. Management, vol. 112, pp. 208-219, 2016.
232-239, 2015. [165] W.-C. Yeh, Y.-M. Yeh, P.-C. Chang, Y.-C. Ke, and V.
[152] C. Wan, Z. Xu, P. Pinson, Z. Y. Dong, and K. P. Wong, Chung, "Forecasting wind power in the Mai Liao Wind Farm
"Probabilistic forecasting of wind power generation using based on the multi-layer perceptron artificial neural network
extreme learning machine," IEEE Transactions on Power model with improved simplified swarm optimization,"
Systems, vol. 29, pp. 1033-1044, 2014. International Journal of Electrical Power & Energy
[153] S. Wu, Y. Wang, and S. Cheng, "Extreme learning machine Systems, vol. 55, pp. 741-748, 2014.
based wind speed estimation and sensorless control for wind [166] R. Rahmani, R. Yusof, M. Seyedmahmoudian, and S.
turbine power generation system," Neurocomputing, vol. Mekhilef, "Hybrid technique of ant colony and particle
102, pp. 163-175, 2013. swarm optimization for short term wind energy forecasting,"
[154] M. Mohandes, S. Rehman, and S. Rahman, "Estimation of Journal of Wind Engineering and Industrial Aerodynamics,
wind speed profile using adaptive neuro-fuzzy inference vol. 123, pp. 163-170, 2013.
system (ANFIS)," Applied Energy, vol. 88, pp. 4024-4032, [167] J. Wang, Y. Wang, and Y. Li, "A Novel Hybrid Strategy
2011. Using Three-Phase Feature Extraction and a Weighted
[155] Y. Bao, H. Wang, and B. Wang, "Short-term wind power Regularized Extreme Learning Machine for Multi-Step
prediction using differential EMD and relevance vector Ahead Wind Speed Prediction," Energies, vol. 11, p. 321,
machine," Neural Computing and Applications, vol. 25, pp. 2018.
283-289, 2014. [168] L. Xiao, X. Xiao, C. Dai, M. Pengy, L. Wang, and H. V.
[156] A. U. Haque, P. Mandal, J. Meng, and M. Negnevitsky, Poor, "Reinforcement Learning-based Energy Trading for
"Wind speed forecast model for wind farm based on a hybrid Microgrids," arXiv preprint arXiv:1801.06285, 2018.
machine learning algorithm," International Journal of [169] L. Xiao, N. B. Mandayam, and H. V. Poor, "Prospect
Sustainable Energy, vol. 34, pp. 38-51, 2015. theoretic analysis of energy exchange among microgrids,"
[157] Y. Liu, J. Shi, Y. Yang, and W.-J. Lee, "Short-term wind- IEEE Transactions on Smart Grid, vol. 6, pp. 63-72, 2015.
power prediction based on wavelet transform–support vector [170] Grid integration of large-capacity Renewable Energy
machine and statistic-characteristics analysis," IEEE sources and use of large-capacity Electrical Energy Storage
Transactions on Industry Applications, vol. 48, pp. 1136- (White Paper): International Electrotechnical Commission
1141, 2012. (IEC).
[158] I. Colak, S. Sagiroglu, and M. Yesilbudak, "Data mining and [171] G. Masson, S. Orlandi, and M. Rekinge, "Global Market
wind power prediction: A literature review," Renewable Outlook For Photovoltaics 2014–2018," European
Energy, vol. 46, pp. 241-247, 2012. Photovoltaic Industry Association.
[159] K. Leahy, R. L. Hu, I. C. Konstantakopoulos, C. J. Spanos, [172] R. Marquez and C. F. Coimbra, "Forecasting of global and
and A. M. Agogino, "Diagnosing wind turbine faults using direct solar irradiance using stochastic learning methods,
machine learning techniques applied to operational data," in ground experiments and the NWS database," Solar Energy,
Prognostics and Health Management (ICPHM), 2016 IEEE vol. 85, pp. 746-756, 2011.
International Conference on, 2016, pp. 1-8. [173] L. Martín, L. F. Zarzalejo, J. Polo, A. Navarro, R.
[160] A. Marvuglia and A. Messineo, "Monitoring of wind farms’ Marchante, and M. Cony, "Prediction of global solar
power curves using machine learning techniques," Applied irradiance based on time series analysis: Application to solar
Energy, vol. 98, pp. 574-583, 2012. thermal power plants energy production planning," Solar
[161] S. Fan, J. R. Liao, R. Yokoyama, L. Chen, and W.-J. Lee, Energy, vol. 84, pp. 1772-1781, 2010.
"Forecasting the wind generation using a two-stage network [174] D. Picault, B. Raison, S. Bacha, J. De La Casa, and J.
based on meteorological information," IEEE Transactions Aguilera, "Forecasting photovoltaic array power production
on Energy Conversion, vol. 24, pp. 474-482, 2009. subject to mismatch losses," Solar Energy, vol. 84, pp. 1301-
[162] D. Lee and R. Baldick, "Short-term wind power ensemble 1309, 2010.
prediction based on Gaussian processes and neural [175] C. W. Chow, B. Urquhart, M. Lave, A. Dominguez, J.
networks," IEEE Transactions on Smart Grid, vol. 5, pp. Kleissl, J. Shields, et al., "Intra-hour forecasting with a total
501-510, 2014. sky imager at the UC San Diego solar energy testbed," Solar
[163] S. Salcedo-Sanz, A. Pastor-Sánchez, L. Prieto, A. Blanco- Energy, vol. 85, pp. 2881-2893, 2011.
Aguilera, and R. García-Herrera, "Feature selection in wind [176] J. Lee and G. Park, "Climate effect analysis on solar energy
speed prediction systems based on a hybrid coral reefs generation in Jeju City," Under review in ARPN Journal of
optimization–Extreme learning machine approach," Energy Engineering and Applied Science, 2016.
Conversion and Management, vol. 87, pp. 10-18, 2014.

VOLUME XX, 2017 9

[177] Y. Gala, Á. Fernández, J. Díaz, and J. R. Dorronsoro, [190] I. Andrea, C. Chrysostomou, and G. Hadjichristofi, "Internet
"Hybrid machine learning forecasting of solar radiation of Things: Security vulnerabilities and challenges," in
values," Neurocomputing, vol. 176, pp. 48-59, 2016. Computers and Communication (ISCC), 2015 IEEE
[178] C. Voyant, G. Notton, S. Kalogirou, M.-L. Nivet, C. Paoli, Symposium on, 2015, pp. 180-187.
F. Motte, et al., "Machine learning methods for solar [191] C. McLellan. (2016, April 9). Cybersecurity predictions for
radiation forecasting: A review," Renewable Energy, vol. 2016: How are they doing? Available:
105, pp. 569-582, 2017. https://www.zdnet.com/article/ransomware-is-now-big-
[179] A. Chaouachi, R. M. Kamel, and K. Nagasaka, "A novel business-on-the-dark-web-and-malware-developers-are-
multi-model neuro-fuzzy-based MPPT for three-phase grid- cashing-in/
connected photovoltaic system," Solar energy, vol. 84, pp. [192] C. McLellan. (2017, April 9). Cybersecurity in an IoT and
2219-2229, 2010. mobile world: The key trends. Available:
[180] A. Chaouachi, R. M. Kamel, R. Andoulsi, and K. Nagasaka, https://www.zdnet.com/article/cybersecurity-in-an-iot-and-
"Multiobjective intelligent energy management for a mobile-world-the-key-trends/
microgrid," IEEE Transactions on Industrial Electronics, [193] A. Anwar and A. N. Mahmood, "Cyber security of smart
vol. 60, pp. 1688-1699, 2013. grid infrastructure," arXiv preprint arXiv:1401.3936, 2014.
[181] H.-T. Yang, C.-M. Huang, Y.-C. Huang, and Y.-S. Pai, "A [194] "Smart Grid Cyber Security," EURELECTRIC
weather-based hybrid method for 1-day ahead hourly D/2016/12.105/64, December 2016.
forecasting of PV power output," IEEE transactions on [195] S. G. I. P. C. S. W. Group, "Introduction to NISTIR 7628
sustainable energy, vol. 5, pp. 917-926, 2014. guidelines for smart grid cyber security," NIST Special
[182] M. R. Hossain, A. M. T. Oo, and A. Ali, "The combined Publication, vol. 154, 2010.
effect of applying feature selection and parameter [196] R. Deng, G. Xiao, R. Lu, H. Liang, and A. V. Vasilakos,
optimization on machine learning techniques for solar Power "False data injection on state estimation in power systems—
prediction," American Journal of Energy Research, vol. 1, Attacks, impacts, and defense: A survey," IEEE
pp. 7-16, 2013. Transactions on Industrial Informatics, vol. 13, pp. 411-423,
[183] J. Li, J. K. Ward, J. Tong, L. Collins, and G. Platt, "Machine 2017.
learning for solar irradiance forecasting of photovoltaic [197] S. Ahmed, Y. Lee, S.-H. Hyun, and I. Koo, "Covert Cyber
system," Renewable Energy, vol. 90, pp. 542-553, 2016. Assault Detection in Smart Grid Networks Utilizing Feature
[184] S. Salcedo-Sanz, C. Casanova-Mateo, A. Pastor-Sánchez, Selection and Euclidean Distance-Based Machine Learning,"
and M. Sánchez-Girón, "Daily global solar radiation Applied Sciences, vol. 8, p. 772, 2018.
prediction based on a hybrid Coral Reefs Optimization– [198] L. Xie, Y. Mo, and B. Sinopoli, "Integrity data attacks in
Extreme Learning Machine approach," Solar Energy, vol. power market operations," IEEE Transactions on Smart
105, pp. 91-98, 2014. Grid, vol. 2, pp. 659-666, 2011.
[185] S. Salcedo-Sanz, C. Casanova-Mateo, J. Muñoz-Marí, and [199] J. Hao, R. J. Piechocki, D. Kaleshi, W. H. Chin, and Z. Fan,
G. Camps-Valls, "Prediction of daily global solar irradiation "Sparse malicious false data injection attacks and defense
using temporal Gaussian processes," IEEE Geoscience and mechanisms in smart grids," IEEE Transactions on
Remote Sensing Letters, vol. 11, pp. 1936-1940, 2014. Industrial Informatics, vol. 11, pp. 1-12, 2015.
[186] Z. Li, S. Rahman, R. Vega, and B. Dong, "A hierarchical [200] Y. Wang, M. M. Amin, J. Fu, and H. B. Moussa, "A Novel
approach using machine learning methods in solar Data Analytical Approach for False Data Injection Cyber-
photovoltaic energy production forecasting," Energies, vol. Physical Attack Mitigation in Smart Grids," IEEE Access,
9, p. 55, 2016. vol. 5, pp. 26022-26033, 2017.
[187] Q. Liu, P. Li, W. Zhao, W. Cai, S. Yu, and V. C. Leung, "A [201] S. Ahmed, Y. Lee, S.-H. Hyun, and I. Koo, "Feature
survey on security threats and defensive techniques of Selection–Based Detection of Covert Cyber Deception
machine learning: a data driven view," IEEE access, vol. 6, Assaults in Smart Grid Communications Networks using
pp. 12103-12117, 2018. Machine Learning," IEEE Access, 2018.
[188] S. Tan, D. De, W.-Z. Song, J. Yang, and S. K. Das, "Survey [202] L. Wei, L. P. Rondon, A. Moghadasi, and A. I. Sarwat,
of security advances in smart grid: A data driven approach," "Review of Cyber-Physical Attacks and Counter Defense
IEEE Communications Surveys & Tutorials, vol. 19, pp. Mechanisms for Advanced Metering Infrastructure in Smart
397-422, 2017. Grid," arXiv preprint arXiv:1805.07422, 2018.
[189] Z. Fan, P. Kulkarni, S. Gormus, C. Efthymiou, G. [203] J. Abdella and K. Shuaib, Peer to Peer Distributed Energy
Kalogridis, M. Sooriyabandara, et al., "Smart Grid Trading in Smart Grids: A Survey vol. 11, 2018.
Communications: Overview of Research Challenges, [204] D. Tellbach and Y.-F. Li, "Cyber-Attacks on Smart Meters
Solutions, and Standardization Activities," IEEE in Household Nanogrid: Modeling, Simulation and
Communications Surveys & Tutorials, vol. 15, pp. 21-38, Analysis," Energies, vol. 11, p. 316, 2018.
2013. [205] Y. Liu, P. Ning, and M. K. Reiter, "False data injection
attacks against state estimation in electric power grids,"

VOLUME XX, 2017 9

ACM Transactions on Information and System Security [219] S. L. T. Ghazi. (2011, April 9). The Increasing Importance
(TISSEC), vol. 14, p. 13, 2011. of Security for the Smart Grid. Available:
[206] S. Paul and Z. Ni, "Vulnerability analysis for simultaneous https://www.elp.com/articles/powergrid_international/print/v
attack in smart grid security," in Power & Energy Society olume-16/issue-4/features/the-increasing-importance-of-
Innovative Smart Grid Technologies Conference (ISGT), security-for-the-smart-grid.html
2017 IEEE, 2017, pp. 1-5. [220] Y. Yuan, Z. Li, and K. Ren, "Quantitative analysis of load
[207] S. Poudel, Z. Ni, and W. Sun, "Electrical distance approach redistribution attacks in power systems," IEEE Transactions
for searching vulnerable branches during contingencies," on Parallel and Distributed Systems, vol. 23, pp. 1731-1738,
IEEE Transactions on Smart Grid, 2016. 2012.
[208] J. Pagliery. (2014, April 8). Hackers attacked the U.S. [221] E. Pallotti and F. Mangiatordi, "Smart grid cyber security
energy grid 79 times this year. Available: requirements," in Environment and Electrical Engineering
http://money.cnn.com/2014/11/18/technology/security/energ (EEEIC), 2011 10th International Conference on, 2011, pp.
y-grid-hack/index.html 1-4.
[209] S. M. Amin and A. M. Giacomoni, "Smart grid, safe grid," [222] T. Sommestad, G. N. Ericsson, and J. Nordlander, "SCADA
IEEE power and energy magazine, vol. 10, pp. 33-40, 2012. system cyber security—A comparison of standards," in
[210] J. Pyper. (2018, April 8). DOE Forms a New Office Power and Energy Society General Meeting, 2010 IEEE,
Dedicated to ‘Energy Infrastructure Security’ Available: 2010, pp. 1-8.
https://www.greentechmedia.com/articles/read/doe-new- [223] Q. Chen, D. Kaleshi, Z. Fan, and S. Armour, "Impact of
office-energy-infrastructure-security- Smart Metering Data Aggregation on Distribution System
cybersecurity#gs.QxgPOh0 State Estimation," IEEE Trans. Industrial Informatics, vol.
[211] J. Desjardins. (2017, April 8). These Are the Countries Most 12, pp. 1426-1437, 2016.
(and Least) Prepared for Cyber Attacks. Available: [224] M. A. Mustafa, N. Zhang, G. Kalogridis, and Z. Fan,
http://www.visualcapitalist.com/countries-least-prepared- "DEP2SA: A decentralized efficient privacy-preserving and
cyber-attacks/ selective aggregation scheme in advanced metering
[212] J. Santiago. (2015, April 8). Top countries best prepared infrastructure," IEEE Access, vol. 3, pp. 2828-2846, 2015.
against cyberattacks. Available: [225] K. Zhou and S. Yang, "A framework of service-oriented
https://www.weforum.org/agenda/2015/07/top-countries- operation model of China‫ ׳‬s power system," Renewable and
best-prepared-against-cyberattacks/ Sustainable Energy Reviews, vol. 50, pp. 719-725, 2015.
[213] A. B. N. Perlroth. (2018, April 8). A Cyberattack Hobbles [226] B. Bitzer and E. S. Gebretsadik, "Cloud computing
Atlanta, and Security Experts Shudder. Available: framework for smart grid applications," in Power
https://www.nytimes.com/2018/03/27/us/cyberattack- Engineering Conference (UPEC), 2013 48th International
atlanta-ransomware.html Universities', 2013, pp. 1-5.
[214] R. McMillan. (2010, April 10). Siemens: Stuxnet worm hit [227] H. Kim, Y.-J. Kim, K. Yang, and M. Thottan, "Cloud-based
industrial systems. Available: demand response for smart grid: Architecture and distributed
https://www.computerworld.com/article/2515570/network- algorithms," in Smart Grid Communications
security/siemens--stuxnet-worm-hit-industrial-systems.html (SmartGridComm), 2011 IEEE International Conference on,
[215] U.-C. P. S. O. T. Force, S. Abraham, H. Dhaliwal, R. J. 2011, pp. 398-403.
Efford, L. J. Keen, A. McLellan, et al., Final report on the [228] X. Zhang, Z. Y. Dong, Z. Wang, C. Xiao, and F. Luo,
august 14, 2003 blackout in the united states and canada: "Quantum cryptography based cyber-physical security
Causes and recommendations: US-Canada Power System technology for smart grids," 2015.
Outage Task Force, 2004. [229] I. O. f. S. (ISO), "ISO/IEC 27001:2013 Information
[216] R. M. L. M. J. A. T. Conway, "Analysis of the Cyber Attack technology — Security techniques — Information security
on the Ukrainian Power Grid," Electricity Information management systems — Requirements," ed: ISO/IEC JTC
Sharing and Analysis CenterMarch 18 2016. 1/SC 27, 2013.
[217] L. Dearden. (2017, April 10). Ukraine cyber attack: Chaos [230] (April 8). CISQ Consortium for IT Software Quality.
as national bank, state power provider and airport hit by Available: http://it-cisq.org/standards/
hackers. Available: [231] (April 8). ISF Information Security Forum. Available:
https://www.independent.co.uk/news/world/europe/ukraine- https://www.securityforum.org/
cyber-attack-hackers-national-bank-state-power-company- [232] "Information Security Forum releases the Standard of Good
airport-rozenko-pavlo-cabinet-a7810471.html Practice for Information Security 2016," ed: Information
[218] G. Kalogridis, M. Sooriyabandara, Z. Fan, and M. A. Security Forum Limited 2016.
Mustafa, "Toward Unified Security and Privacy Protection [233] (April 8). NERC North American Electric Reliability
for Smart Meter Networks," IEEE Systems Journal, vol. 8, Corporation. Available:
pp. 641-654, 2014. https://www.nerc.com/Pages/default.aspx

VOLUME XX, 2017 9

[234] (2014, April 8). Framework for Improving Critical [252] S. Li, Y. Yılmaz, and X. Wang, "Quickest detection of false
Infrastructure Cybersecurity (1.0 ed.). Available: data injection attack in wide-area smart grids," IEEE
https://www.nist.gov/sites/default/files/documents/cyberfram Transactions on Smart Grid, vol. 6, pp. 2725-2735, 2015.
ework/cybersecurity-framework-021214.pdf [253] S. Pan, T. Morris, and U. Adhikari, "Developing a hybrid
[235] (April 8). Standard ISO/IEC 15408, CC v3.1. Release 4 intrusion detection system using data mining for power
Available: http://commoncriteria.pl/index.php/en/common- systems," IEEE Transactions on Smart Grid, vol. 6, pp.
criteria-standard/standard-documentation/standard-iso-iec- 3104-3113, 2015.
15408-cc-v3-1-release-4 [254] M. Ozay, I. Esnaola, F. T. Y. Vural, S. R. Kulkarni, and H.
[236] (April 8). Common Criteria. Available: V. Poor, "Machine learning methods for attack detection in
https://www.commoncriteriaportal.org/ the smart grid," IEEE transactions on neural networks and
[237] B. Fraser, "RFC 2196. Site Security Handbook. 1997," URL: learning systems, vol. 27, pp. 1773-1786, 2016.
http://www. faqs. org/rfcs/rfc2196. html, 2003. [255] Y. Xin, L. Kong, Z. Liu, Y. Chen, Y. Li, H. Zhu, et al.,
[238] T. I. S. o. Automation. (2012, April 8). Applying ISA IEC "Machine Learning and Deep Learning Methods for
62443 to Control Systems - ISA Automation Week 2012: Cybersecurity," IEEE Access, 2018.
Security. Available: https://www.isa.org/store/applying-isa- [256] V. B. Krishna, G. A. Weaver, and W. H. Sanders, "PCA-
iec-62443-to-control-systems-isa-automation-week-2012- based method for detecting integrity attacks on advanced
security/122496 metering infrastructure," in International Conference on
[239] M. Dworkin, "Recommendation for block cipher modes of Quantitative Evaluation of Systems, 2015, pp. 70-85.
operation. methods and techniques," NATIONAL INST OF [257] S. Amin, G. A. Schwartz, A. A. Cardenas, and S. S. Sastry,
STANDARDS AND TECHNOLOGY GAITHERSBURG "Game-theoretic models of electricity theft detection in
MD COMPUTER SECURITY DIV2001. smart utility networks: Providing new capabilities with
[240] E. Barker, "SP 800-67 Rev. 2, Recommendation for Triple advanced metering infrastructure," IEEE Control Systems,
Data Encryption Algorithm (TDEA) Block Cipher," NIST vol. 35, pp. 66-81, 2015.
special publication, vol. 800, p. 67, 2017.
[241] E. B. Barker, "Digital Signature Standard (DSS)," 2009.
[242] Q. H. Dang, "Secure hash standard," 2015.
[243] M. J. Dworkin, "Recommendation for block cipher modes of
operation: The CMAC mode for authentication," 2016.
[244] M. J. Dworkin and M. Dworkin, Recommendation for block EKLAS HOSSAIN (M’09, SM’17)
cipher modes of operation: The CCM mode for received his PhD from the College of
authentication and confidentiality, 2007. Engineering and Applied Science at
University of Wisconsin Milwaukee
[245] M. J. Dworkin, "Recommendation for block cipher modes of (UWM) in 2016. He received his MS in
operation: Galois/Counter Mode (GCM) and GMAC," 2007. Mechatronics and Robotics Engineering
[246] J. M. Turner, "The keyed-hash message authentication code from International Islamic University of
Malaysia, Malaysia in 2010 and BS in
(hmac)," Federal Information Processing Standards Electrical & Electronic Engineering from
Publication, 2008. Khulna University of Engineering and
[247] J. Baek, Q. H. Vu, J. K. Liu, X. Huang, and Y. Xiang, "A Technology, Bangladesh in 2006. Dr.
Hossain has been working in the area of distributed power systems and
secure cloud computing based framework for big data
renewable energy integration for last ten years and he has published a
information management of smart grid," IEEE transactions number of research papers and posters in this field. He is now involved
on cloud computing, vol. 3, pp. 233-244, 2015. with several research projects on renewable energy and grid tied microgrid
[248] A. L. Buczak and E. Guven, "A survey of data mining and system at Oregon Tech, as an Assistant Professor in the Department of
Electrical Engineering and Renewable Energy since 2015. He is currently
machine learning methods for cyber security intrusion serving as an Associate Editor of IEEE Access. His research interests
detection," IEEE Communications Surveys & Tutorials, vol. include modeling, analysis, design, and control of power electronic
18, pp. 1153-1176, 2016. devices; energy storage systems; renewable energy sources; integration of
distributed generation systems; microgrid and smart grid applications;
[249] M. H. Bhuyan, D. K. Bhattacharyya, and J. K. Kalita, robotics, and advanced control system. Dr. Hossain, with his dedicated
"Network anomaly detection: methods, systems and tools," research team, is looking forward to explore methods to make the electric
Ieee communications surveys & tutorials, vol. 16, pp. 303- power systems more sustainable, cost-effective and secure through
extensive research and analysis on energy storage, microgrid system and
336, 2014. renewable energy sources.
[250] J. Singh and M. J. Nene, "A survey on machine learning
techniques for intrusion detection systems," International IMTIAJ KHAN received the B.Sc. degree
in Electrical and Electronic Engineering
Journal of Advanced Research in Computer and
from Bangladesh University of
Communication Engineering, vol. 2, pp. 4349-4355, 2013. Engineering and Technology (BUET) in
[251] O. Kosut, L. Jia, R. J. Thomas, and L. Tong, "Malicious data 2017. His undergraduate thesis was on
attacks on the smart grid," IEEE Transactions on Smart “Optimization and Comparison of Single
Walled and Double Walled Carbon
Grid, vol. 2, pp. 645-658, 2011. Nanotube Field Effect Transistors

VOLUME XX, 2017 9

(CNTFETs)”. His previous works were presented in IEEE NANO 2017

and TENCON 2017. His research interests include Nanotechnology,
Plasmonics, Photonics, and Smart Grid. He is currently working on Power
Quality Improvement and Noble Manipulation on Particles by Optical
Force.

FUAD UN-NOOR received his B.Sc. from

Khulna University of Engineering and
Technology (KUET) in Electrical and
Electronic Engineering in 2017. His previous
works focused on electric vehicles and
power systems. He is currently working on
smart grid/microgrid, and energy storage
systems. His research interests include
electric vehicles, power and energy systems,
renewable energy systems, smart
grid/microgrid, and energy storage systems.

SARDAR SHAZALI SIKANDER (M’12–

SM’15) received B.S. degree in Electronics
from the International Islamic University
Islamabad, in 2012. Afterwards he has
received M.S. degree in Electrical
Engineering from National University of
Science and Technology Islamabad in 2014.
His research interests include robotics,
power electronics converters, nonlinear
control and fuzzy logic control.

MD. SAMIUL HAQUE SUNNY received

the B.Sc. degree in Electrical and Electronic
Engineering from Khulna University of
Engineering and Technology (KUET) in
2017. His undergraduate thesis was on “High
performance state parameter observation
with sensorless vector control of induction
motor”. His previous works were presented
in ICEEICT 2016, ICAEE 2017, IC2IT 2017,
EICT 2017, and IEEE R10-HTC. His
research interest includes Artificial
Intelligence, Brain Computer Interface,
Digital signal and Image Processing, Data mining, Power System Stability,
and Robotics. He is currently working on Data mining with AI algorithms,
EEG signal for better BCI application and Structures of CNN for
upgrading its performance in image recognition.

VOLUME XX, 2017 9

Training Material
No ratings yet
Training Material
160 pages
11 - 2023 - An Ecological Robustness Oriented Optimal Power Flow For Power Systems' Survivability
No ratings yet
11 - 2023 - An Ecological Robustness Oriented Optimal Power Flow For Power Systems' Survivability
16 pages
Reactive Power Compensation Using Capacitor Banks
100% (1)
Reactive Power Compensation Using Capacitor Banks
82 pages
EEE221
No ratings yet
EEE221
544 pages
Protecciones en DER
No ratings yet
Protecciones en DER
16 pages
Proposal
No ratings yet
Proposal
20 pages
Matlab
100% (1)
Matlab
83 pages
A Historical Review of Robust Control
100% (1)
A Historical Review of Robust Control
4 pages
Matrix Functions Via Jordan Canonical
No ratings yet
Matrix Functions Via Jordan Canonical
10 pages
Solving Optimal Power Flow With Voltage Constraints Using Matlab Optimization Toolbox
50% (2)
Solving Optimal Power Flow With Voltage Constraints Using Matlab Optimization Toolbox
55 pages
These DPetrichenko VF PDF
No ratings yet
These DPetrichenko VF PDF
175 pages
Optimal Power Flows: J Carpentier
100% (1)
Optimal Power Flows: J Carpentier
13 pages
Boundary Load Flow Solutions
No ratings yet
Boundary Load Flow Solutions
75 pages
EE5712 Power System Reliability:: Outage Model & Power Flow Analysis
No ratings yet
EE5712 Power System Reliability:: Outage Model & Power Flow Analysis
69 pages
Lec 7
100% (1)
Lec 7
97 pages
Microgrids Benefits Models Barriers and Suggested Policy Initiatives For The Commonwealth of Massachusetts - pdf1874040306 PDF
No ratings yet
Microgrids Benefits Models Barriers and Suggested Policy Initiatives For The Commonwealth of Massachusetts - pdf1874040306 PDF
66 pages
Electrical Machines - 1
No ratings yet
Electrical Machines - 1
256 pages
Smart Grid
100% (2)
Smart Grid
44 pages
Chapter 1 Cyber Range
No ratings yet
Chapter 1 Cyber Range
20 pages
3 Body
No ratings yet
3 Body
18 pages
New Proposal Final
0% (1)
New Proposal Final
17 pages
Interior Point
No ratings yet
Interior Point
213 pages
Micro Grid
No ratings yet
Micro Grid
10 pages
Control Strategies For Microgrids Black Start and Islanded Operation
No ratings yet
Control Strategies For Microgrids Black Start and Islanded Operation
21 pages
Pes Ed Tut Grid Energy1 072120 SLD
No ratings yet
Pes Ed Tut Grid Energy1 072120 SLD
120 pages
Optimal Power Flow Using Distributed Generation
No ratings yet
Optimal Power Flow Using Distributed Generation
6 pages
Operation of Restructured Power Systems
No ratings yet
Operation of Restructured Power Systems
33 pages
Antenna Azimuth Position Control System Verification
100% (1)
Antenna Azimuth Position Control System Verification
4 pages
Igcse
No ratings yet
Igcse
9 pages
1 - Mark-Ahlstrom - ESIG - The Grid As A Hostage of Its History - The Many Ways That Grid Forming Inverters Will Change Power System
No ratings yet
1 - Mark-Ahlstrom - ESIG - The Grid As A Hostage of Its History - The Many Ways That Grid Forming Inverters Will Change Power System
15 pages
Problems and Prospects of General Insurance in Bangladesh
75% (4)
Problems and Prospects of General Insurance in Bangladesh
56 pages
Deployments and Deployment Attempts: Futuregrid, and The More Modern Intergrid and Intragrid
No ratings yet
Deployments and Deployment Attempts: Futuregrid, and The More Modern Intergrid and Intragrid
22 pages
Renewables and Grid Stability
No ratings yet
Renewables and Grid Stability
15 pages
Day2 - 3 - Hosting Capacity Tools and Examples
No ratings yet
Day2 - 3 - Hosting Capacity Tools and Examples
30 pages
An Analysis of Single Wire Earth Return (SWER) System For Rural e PDF
No ratings yet
An Analysis of Single Wire Earth Return (SWER) System For Rural e PDF
77 pages
Optimum Design of Wind-Diesel Hybrid System For Re PDF
No ratings yet
Optimum Design of Wind-Diesel Hybrid System For Re PDF
9 pages
Hybrid Solar Seminar
No ratings yet
Hybrid Solar Seminar
30 pages
Modeling Analysis and Control of Voltage-Source Converter in Mic
No ratings yet
Modeling Analysis and Control of Voltage-Source Converter in Mic
172 pages
Vulnerability Assesment in Power Systems 225-1258-1-PB
No ratings yet
Vulnerability Assesment in Power Systems 225-1258-1-PB
9 pages
Advanced Inverter Report 2013 Final
No ratings yet
Advanced Inverter Report 2013 Final
16 pages
Adaptive Protection Scheme For Smart Grids
No ratings yet
Adaptive Protection Scheme For Smart Grids
6 pages
Survey of Grid Forming Inverter Applications Julia Matevosyan
No ratings yet
Survey of Grid Forming Inverter Applications Julia Matevosyan
18 pages
Smart Grid - Integration of Distributed Generation Prof Mohan Kolhe
No ratings yet
Smart Grid - Integration of Distributed Generation Prof Mohan Kolhe
20 pages
1 RIYA Immmm
No ratings yet
1 RIYA Immmm
60 pages
Reactive Power Compensation Techniques in Transmission Lines
No ratings yet
Reactive Power Compensation Techniques in Transmission Lines
3 pages
RAC QB Final-2023
No ratings yet
RAC QB Final-2023
9 pages
App Note DC DC Stability V2
No ratings yet
App Note DC DC Stability V2
17 pages
SSP Cakram5 6
No ratings yet
SSP Cakram5 6
420 pages
Introduction To Microgrids - Securicon - 2013 - 1
No ratings yet
Introduction To Microgrids - Securicon - 2013 - 1
13 pages
S71937 - Enabling Intelligent Storage To Process Data For Ai Application Ibm
No ratings yet
S71937 - Enabling Intelligent Storage To Process Data For Ai Application Ibm
21 pages
Source Follower: (Common-Drain Amplifier)
No ratings yet
Source Follower: (Common-Drain Amplifier)
40 pages
Deployment of Microgrids in India
No ratings yet
Deployment of Microgrids in India
34 pages
Project Milk Curdling
No ratings yet
Project Milk Curdling
15 pages
UCSP 1st Q Budget Work
No ratings yet
UCSP 1st Q Budget Work
1 page
Design and Application of Microgrid Operation Control System Based On IEC 61850
No ratings yet
Design and Application of Microgrid Operation Control System Based On IEC 61850
8 pages
Movement in The Classroom
No ratings yet
Movement in The Classroom
7 pages
IEEE Smart Grid Domains and SubDomains Definitions
100% (1)
IEEE Smart Grid Domains and SubDomains Definitions
5 pages
Regulation of Distribution Network Voltage Using Dispersed Battery Storage Systems: A Case Study of A Rural Network
No ratings yet
Regulation of Distribution Network Voltage Using Dispersed Battery Storage Systems: A Case Study of A Rural Network
8 pages
Power Flow Control With FACTS
100% (1)
Power Flow Control With FACTS
18 pages
Smart Grid
No ratings yet
Smart Grid
11 pages
Conti USA IFS Hydraulic Hoses Fittings Catalog 2016
No ratings yet
Conti USA IFS Hydraulic Hoses Fittings Catalog 2016
444 pages
Littering Stats Essay
No ratings yet
Littering Stats Essay
3 pages
NZ Pa 36 New Zealand Numeracy Stages 1 To 8 Weekly Planning Template English Ver 2
No ratings yet
NZ Pa 36 New Zealand Numeracy Stages 1 To 8 Weekly Planning Template English Ver 2
12 pages
The Star Weaver
No ratings yet
The Star Weaver
2 pages
Productores de Banano de Nicaragua Probanic Datos Climáticos de Estación Finca San Luis Enero, 2011
No ratings yet
Productores de Banano de Nicaragua Probanic Datos Climáticos de Estación Finca San Luis Enero, 2011
17 pages
MILP Unit Commitment Formulation
No ratings yet
MILP Unit Commitment Formulation
8 pages
Tax Invoice: 1046.17 Total Invoice Amount Rs
No ratings yet
Tax Invoice: 1046.17 Total Invoice Amount Rs
2 pages
ESD Assignment
No ratings yet
ESD Assignment
14 pages
3.2. Perspectives On Listening Ho
No ratings yet
3.2. Perspectives On Listening Ho
35 pages
Studying The Operation of A Smart Grid With The MyGrid MiniSystem For NI MyDAQ (Piomeer Release)
No ratings yet
Studying The Operation of A Smart Grid With The MyGrid MiniSystem For NI MyDAQ (Piomeer Release)
4 pages
Hybrid Micro-Grids and Their Core Technical Challenges: Abstract
No ratings yet
Hybrid Micro-Grids and Their Core Technical Challenges: Abstract
3 pages
01-Sap Annual Report2023 Ang
No ratings yet
01-Sap Annual Report2023 Ang
119 pages
FSBC01 The Use of Repair and Maintenance Budget For Buildings
No ratings yet
FSBC01 The Use of Repair and Maintenance Budget For Buildings
5 pages
Tutorial DSL TestTroubleshooting
No ratings yet
Tutorial DSL TestTroubleshooting
6 pages
Applications HVAC EN
No ratings yet
Applications HVAC EN
21 pages
Reliability Modeling of Distributed Generation in Conventional Distribution Systems Planning and Analysis
No ratings yet
Reliability Modeling of Distributed Generation in Conventional Distribution Systems Planning and Analysis
6 pages
PDF Living On A Prayer - English Version
No ratings yet
PDF Living On A Prayer - English Version
17 pages
Datasheet - Hexaband APE4518R14v06 PDF
No ratings yet
Datasheet - Hexaband APE4518R14v06 PDF
3 pages
01 Pilot Project Palau
No ratings yet
01 Pilot Project Palau
40 pages
Cbo - Elen4003a - 2023
No ratings yet
Cbo - Elen4003a - 2023
4 pages
1113
No ratings yet
1113
1 page
Impact of HL On QOL
No ratings yet
Impact of HL On QOL
8 pages
Synapse - Test Automation Engineer
No ratings yet
Synapse - Test Automation Engineer
1 page
Prepaid Energy Meter Using Smart Card
No ratings yet
Prepaid Energy Meter Using Smart Card
14 pages
Family Emergency Plan
No ratings yet
Family Emergency Plan
2 pages
Director of Training
No ratings yet
Director of Training
2 pages
Daftar Harga Barang Toko GMC Mojokerto Jl. Gajah Mada No. 42 Tlp. 0321-7229919 Mojokerto
No ratings yet
Daftar Harga Barang Toko GMC Mojokerto Jl. Gajah Mada No. 42 Tlp. 0321-7229919 Mojokerto
6 pages
Advanced Electric Drives: Analysis, Control, and Modeling Using MATLAB / Simulink
From Everand
Advanced Electric Drives: Analysis, Control, and Modeling Using MATLAB / Simulink
Ned Mohan
No ratings yet
Simulation of Some Power System, Control System and Power Electronics Case Studies Using Matlab and PowerWorld Simulator
From Everand
Simulation of Some Power System, Control System and Power Electronics Case Studies Using Matlab and PowerWorld Simulator
Dr. Hedaya Mahmood Alasooly
No ratings yet
Wind Power Generation A Clear and Concise Reference
From Everand
Wind Power Generation A Clear and Concise Reference
Gerardus Blokdyk
No ratings yet

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.

Smart Grid Future PDF

Uploaded by

Smart Grid Future PDF

Uploaded by

This article has been accepted for publication in a future issue of this journal, but has not been

Application of Big Data and Machine Learning

Corresponding author: Eklas Hossain (e-mail: eklas.hossain@oit.edu).

VOLUME XX, 2017 1

LIST OF ABBREVIATIONS CCM Cipher block chaining-message authentication code

Author Name: Preparation of Papers for IEEE Access (February 2017)

environment along with efficient generation and distribution,

Author Name: Preparation of Papers for IEEE Access (February 2017)

Author Name: Preparation of Papers for IEEE Access (February 2017)

(LPRF) communication using a sub-1 GHz mesh network is

Generation Monitoring of energy generation Sabotage

FIGURE 5. System architecture for load shedding and smart load

Author Name: Preparation of Papers for IEEE Access (February 2017)

Author Name: Preparation of Papers for IEEE Access (February 2017)

also have certain limitations, which hinders the rise of a

service provider domain, smart appliances provide

Author Name: Preparation of Papers for IEEE Access (February 2017)

Fayyad Cios SEMMA CRISP-DM

[158]. Machine learning technique has been applied to

Author Name: Preparation of Papers for IEEE Access (February 2017)

VI. CYBER SECURITY IN SMART GRID

A. CYBER-SECURITY CHALLENGES IN SMART GRID

control systems are the other major targets. Information

Name Subscriber Party responsible for the account

Author Name: Preparation of Papers for IEEE Access (February 2017)

Author Name: Preparation of Papers for IEEE Access (February 2017)

FIGURE 13. Attacks on major power grid components during 1994-

FIGURE 14. Likelihood of electric supply interruption from cyber-

B. COUNTER-MEASURES FOR CYBER-THREATS

Author Name: Preparation of Papers for IEEE Access (February 2017)

characteristic uncertainty. This is the fundamental approach

251-253]. Ahmed et al. proposed a machine learning

government initiatives, and their approach to

VOLUME XX, 2017 9

VOLUME XX, 2017 9

VOLUME XX, 2017 9

VOLUME XX, 2017 9

VOLUME XX, 2017 9

VOLUME XX, 2017 9

VOLUME XX, 2017 9

VOLUME XX, 2017 9

VOLUME XX, 2017 9

VOLUME XX, 2017 9

(CNTFETs)”. His previous works were presented in IEEE NANO 2017

FUAD UN-NOOR received his B.Sc. from

SARDAR SHAZALI SIKANDER (M’12–

MD. SAMIUL HAQUE SUNNY received

VOLUME XX, 2017 9

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.