See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/266375284

Recent Advances and Trends of Cyber-Physical Systems and Big Data Analytics
in Industrial Informatics

Conference Paper · July 2014

DOI: 10.13140/2.1.1464.1920

CITATIONS READS

121 18,676

3 authors:

Jay Lee Behrad Bagheri

Foxconn Technology Group (On Leave from Univ. of Cincinnati) Blue Cross and Blue Shield of North Carolina
435 PUBLICATIONS   13,755 CITATIONS    29 PUBLICATIONS   3,431 CITATIONS   

SEE PROFILE SEE PROFILE

Hung-An Kao
University of Cincinnati
23 PUBLICATIONS   4,094 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Prognostics Patents View project

Industrial AI View project

All content following this page was uploaded by Behrad Bagheri on 02 October 2014.

The user has requested enhancement of the downloaded file.

 Proceeding of Int. Conference on Industrial Informatics (INDIN) 2014

Recent Advances and Trends of Cyber-Physical

Systems and Big Data Analytics in Industrial
Informatics

Jay Lee, Behrad Bagheri, Hung-An Kao

Center for Intelligent Maintenance Systems
University of Cincinnati
Cincinnati, United States

Abstract— In today’s competitive business environment, II. RECENT ADVANCES AND TERMINOLOGIES
companies are facing challenges in dealing with big data issues
for rapid decision making for improved productivity. Many In such competitive and creative environment, new terms
manufacturing systems are not ready to manage big data due to and phrases have born to address the current requirement and
the lack of smart analytics tools. Germany is leading a demands of the industry. Recently, Germany has announced
transformation toward 4th Generation Industrial Revolution the Industry 4.0 methodology as the fourth industrial
(Industry 4.0) based on Cyber-Physical System based revolution. Three past revolutions are "First mechanical loom"
manufacturing and service innovation. As more software and in 1784, "First assembly line" in 1870, "First programmable
embedded intelligence are integrated in industrial products and logic controller (PLC)" in 1969 respectively. Based on the
systems, predictive technologies can further intertwine intelligent industry 4.0 terminology, intelligent data analysis and
algorithms with electronics and tether-free intelligence to predict interconnected systems are combined together to generate a
product performance degradation and autonomously manage brand new aspect in factory transformation and production
and optimize product service needs. This article addresses the management. Table 1 represents the differences between a
trends of industrial transformation in big data environment as today's factory and an Industry 4.0 factory. In current industry
well as the readiness of smart predictive informatics tools to environment, providing high-end quality service or product
manage big data to achieve transparency and productivity.
with the least cost is the key to success and industrial factories
are trying to achieve as much performance as possible to
Keywords—Industry 4.0; Cyber Physical Systems; Prognostics
and Health Management; Big Data; increase their profit as well as their reputation. In this way,
various data sources are available to provide worthwhile
I. INTRODUCTION information about different aspects of the factory. In this stage,
the utilization of data for understanding the current condition
In the past decade, emergence of promising tools such as
and detect faults and failures is an important topic to research.
Enterprise Systems provided companies with solutions to
e. g. in production, there are various commercial tools
improve their productivity and service quality. But today's
available to provide OEE (Overall Equipment Effectiveness)
competitive nature of world industry enforces companies to
information to factory management in order to highlight root
implement more recent technologies to secure their position
cause of problems and possible faults in the system. In
among competitors [1]. Recent technological advances in the
contrast, in an Industry 4.0 factory, in addition to condition
field of communication and computer science have provided
monitoring and fault diagnosis, components and systems are
cost-effective solutions for companies to acquire and transfer
able to gain self-awareness and self-predictiveness, which will
gigantic amount of data from their fleet of assets. Consequently
provide management with more insight on the status of the
Handling these huge sets of data is not easily achievable,
factory. Furthermore, peer-to-peer comparison and fusion of
therefore supporting "Big Data" is one of the most recent topics
health information from various components provides a precise
in the world industry. Introduction of methods and terms such
health prediction in component and system levels and enforce
as internet of things and interconnected systems are among the
factory management to trigger required maintenance at the best
efforts of researchers and industrial companies to address
possible time to reach just-in time maintenance and gain near
applicable solutions in the "Big Data" environment. Therefore,
zero downtime. In addition to Industry 4.0, Cyber-physical
the requirement of systematic approaches to handle and
system (CPS) is a phrase representing the integrated
analyze enterprise data in the Big Data environment is the
computational and physical capabilities such as sensing,
purpose of several research studies including current paper. [2].
communication and actuation to physical world[3], which is
addressed by American government since 2007 as a new
developments strategy[4],[5].
 Proceeding of Int. Conference on Industrial Informatics (INDIN) 2014

Table 1: Comparison of today’s factory and an Industry 4.0 factory

Applications of CPS include but not limited to serious downtimes in the factory floor and PHM as an ever-
manufacturing, Secure Control, medical devices, growing pioneer research domain provides useful approaches
Environmental Control, aviation, advanced automotive to detect those invisible problems and avoid unplanned
systems, process control, energy control, traffic control and downtimes to overwhelm uncertainties in the systems.
safety, smart structures and so on[6]. Having in mind that CPS Remaining useful life prediction, fault diagnosis and fault
is in its immature stage and it covers a broad range of scientific detection are among the useful tools provided by PHM
era, interactive collaborations between industry and academia, algorithms for bringing transparency in every aspect of future
especially long-term collaborations such as the tasks that have industry by analyzing sensory and system level data. In Past
been done in Advanced Research and Technology for few years, different aspects of industry as well as different
Embedded Intelligence Systems (ARTEMIS) in Europe or machines or components have been targeted by researchers for
Center for Intelligent Maintenance Systems (IMS) in United developing reliable PHM solutions. Aircraft engines [8],
States will help in identification of challenges and significantly industrial robots [9], machine tools [10], electrical motors[11],
accelerate the progress of CPS[4], [7]. wind turbine[12], batteries [13], gearboxes [14], bearings [15],
pumps [16] and etc are only few examples of a broad range of
In the implementation of advanced terms such as Industry assets.
4.0 and Cyber Physical Systems, sole presence of connectivity
between machines and using sensors is not useful. To leverage IV. INTEGRATED SYSTEMS
these advanced technologies, correct information has to be
present at the right time for the right purpose. In this situation, In previous sections, we introduced Industry 4.0 and Cyber-
a 6C system that is consist of Connection (sensor and Physical Systems as two leading infrastructures for managing
networks), Cloud (data on demand), Cyber (model & memory), data and leaning toward more efficient production in current
Content (meaning and correlation), Community (sharing & industry. In addition, we presented prognostics and health
collaboration), and Customization (personalization and value) management methods as powerful and applicable approaches
[7] can enhance the information system. In this scenario and in for detecting invisible faults and failures among factories fleet.
order to provide useful insight to the factory management and In current section, advantages of implementing PHM
gain correct content, data has to be processed with advanced algorithms in Industry 4.0 and CPS application will be
tools to generate relative information. Considering the presence presented and the next section introduces a methodology for
of visible and invisible issues in an industrial factory, the implementing cyber-physical systems in industry 4.0
information generation algorithm has to capable of detecting applications.
and addressing invisible issues such as machine degradation, Today, most PHM methods have only access to target asset
component wear, etc in the factory floor. data. Asset data might contain hundreds of sensory and system
data readings and provide good insight and prognostics results
III. PROGNOSTICS AND HEALTH MANAGEMENT for the current asset. But, there are much more sources of
Prognostics and health management (PHM) is an enrich information that are not considered in today's implementation
research area that mostly deals with component wear and of PHM methods. Peer-to-peer evaluation through fleet of
degradation. Factory wide transparency is one of the most assets, life-cycle historical data from identical assets and
important targets of PHM because in contrast with visible system configuration are few examples of the huge portion of
issues, invisible issues might happen due machine degradation, data that is ignored by now.. For example, in PHM, various
component wear and etc. while operators and factory managers failure signatures are required for accurate fault diagnosis of
are not aware of them. Indentifying those issues can lead to assets and improve the reliability of maintenance scheme.
 Proceeding of Int. Conference on Industrial Informatics (INDIN) 2014

Fig. 1. The general framework of the application of PHM algorithms in detecting invisible issues in industry using Watchdog Agent® tool

Acquiring every possible failure signature from one instance of method has to be broad enough to truly leverage all the
assets is not likely to happen but in an interconnected system advantages of cyber-physical systems.
which manages hundreds of similar assets among the fleet, the
In this article, we propose “Time Machine Methodology for
occurrence likelihood of various failures much higher.
Cyber-Physical Systems” which is in charge of perfectly
Therefore, the PHM analytical engine can capture those failure
signatures from various assets to refine its failure detection and organize available data in Big Data environment to be prepared
for usage in PHM algorithms. Every single component of the
estimation capabilities. This huge improvement is possible by
utilizing the concept of cyber-physical systems for PHM fleet will have a representative Time Machine record in the
cyber space. This cyber delegate extracts worthwhile
applications in which a cyber twin model (avatar) of real
machine is created and operated in cloud information from the pool of available data and normalizes it
for further analysis. Extracted information include but not
platform parallel to actual asset. Leveraging the Integrated
knowledge available in cyber physical model, health condition limited to implementation history, stress and load, operation
parameters, system configurations and maintenance records.
of assets can be accurately simulated and appropriately
presented to dedicated users upon their demand without Once the actual component is failed, it will be removed from
the parent machine and will no longer exist to participate in
geographical limitations. Furthermore, the interconnectivity of
fleet of identical assets to their cyber-physical model not only analytics. But its cyber twin (Time machine record) will
remain without any time constraints. Unlimited existence of
provides the opportunity of peer-to-peer evaluation and
prognostic library accumulation but also enables PHM cyber twins results in continuous accumulation of Time
Machine records and consequently gathering various operation
algorithms to have access to various life stages of different
assets as well as test stage data. parameters from broad range of identical components.
Normalization of parameters, in which further research efforts
V. METHODOLOGY FOR DESIGNING CPS BASED have to be conducted, ensures the comparability of time
INDUSTRY 4.0 SYSTEMS machine records with each other for identical components.
Additionally, time machine records obey the hierarchical
Knowing the capabilities of cyber-physical systems, a relation of actual components as well and every cyber twin has
promising methodology for designing Industry 4.0 applications access to records of its predecessor and ancestor components.
based on Cyber-Physical Systems can be developed. As it was Such information rich environment brings significant
discussed in previous section, interconnectivity provides access robustness to PHM algorithms for continuous and accurate
to vast amount of data. But, sole availability of data does not predicting and monitoring of the factory. Ultimately, this
create a significant advantage. Therefore, an adaptive yet methodology brings ultimate implementation of cyber physical
powerful methodology is required to manage, categorize and system into action for designing an Industry 4.0 factory.
process data for further analysis by PHM algorithms. This
 Proceeding of Int. Conference on Industrial Informatics (INDIN) 2014

Fig. 2. Cyber-Physical Model for Industrial Robot Monitoring

VI. CASE STUDIES system by using torque and speed data. Due to its non-invasive
nature torque monitoring is a popular fault detection method
A. Industrial Robot for monitoring health condition of industrial robots and
In this section we consider the development of a cyber therefore, most of the research efforts in this area have been
physical system for health monitoring of industrial robots. Fig. focused on this parameter. In addition, nonlinear relation
2. shows the schematic view of the cyber-physical model for between operating speed and torque cast a challenge on PHM
current case study on industrial robots. This study was focused algorithms to correctly determine the health state of the robot.
on preparing a predictive health monitoring solution for a fleet In addition to condition data (torque and speed) the cyber
of 30 industrial robots in production line. Due to variety of line physical model obtains various configuration parameters such
speeds, a more complicated multi-regime approach had to be as gear ratio, load ratio, pressure calibration, type of tooling for
undertaken for establishing a robust prognostics and health robot servo guns and assigned products to specific robots from
management algorithm as the core analytics of cyber physical production line.

Fig. 3. Methodology for Designing CPS based Industry 4.0 Systems

 Proceeding of Int. Conference on Industrial Informatics (INDIN) 2014
These configuration parameters help the core of the model to 2. Adequate visualization system to deliver the right
standardize and adaptively cluster the operation data for more information to the right people. Some information that requires
accurate processing. Ultimately, the entire analytical engine of the immediate intervention of the driver may be displayed for
this study was set and established on the cloud where the whole the actual user (driver) on the dashboard, while other forms of
dataset (condition and configuration data) were stored on the information (including more detailed diagnostic information)
cloud storage and health monitoring algorithm used the data would be necessary for a maintenance personnel or logistics
center and used to schedule maintenance or replacements.
stored on the cloud (from entire fleet) to calculate the health
3. Enabling tether-free communication to capture this
condition of each individual instance. The outcome of PHM
information in real-time from the vehicle and deliver it to a
algorithms were presented to users as enriched infographics central logistics center that will advise the driver about his
through a web-based user interface battery condition and could use GPS trip information to direct
B. Virtual Battery the driver to the nearest service center or rapid charging center
on his trip.
The battery pack is one of the most significant components
4. Simulating and predicting battery behavior such as
of an Electric or Hybrid Vehicle. The uncertainty in the driving
SoC (State of Charge) and SoH (State of Health) in other
range, the batteries reliability and service life, and battery
operation conditions based on external parameters(
safety concerns, are all challenges that must be overcome in
environment , driving style, etc ) and internal parameters (
order to achieve more widespread adoption of electric vehicles.
battery type, and battery age) without take a long time to test a
In order to meet the challenges and rising demand in hybrid
battery in that condition.
and electric vehicles in recent years, the importance of batteries
5. Help the designers to find the flaws or design issues
has become more and more important[17]. Therefore, fully
for a better hybrid cycle management in parallel with a better
understanding of the dynamic performance of batteries under
estimation of the selected pack’s lifetime expectancy in the
various conditions is of significant importance. However,
specific application.
batteries operation condition includes several stress factors
6. Help the battery manufacturers to locate the faulty
such as environment temperature, humidity, driving style,
packs and study them to refine their design, manufacturing
charging level, discharge rate and road conditions. To achieve
process or raw material selection.
this battery understanding a battery model is needed which
7. Integration of the wireless solutions into the
should provide capabilities for health evaluation and failure
prognostic tool could also open an opportunity for better
prediction, through simulation of cell performance under
monitoring of the vehicle during standard or extended warranty
different conditions. Additionally, using such a battery model,
period. The faulty system could be recalled for repair long
the functionality of a manufactured battery can be fed back to
before the usual inspection periods. This in turn can reduce the
design and manufacturing suppliers to recognize the potential
costs of warranty that could rise from such occasions. There is
impact of design and enhancement of manufacturing process
also a possibility to integrate this system into engine
on battery performance[18].
management sensors. This will enable the designers to study
An overview of the electric vehicle and battery health the battery aging process in different climatic conditions that
management and prognostic platform is illustrated in fig. 4, the vehicle operates in. This over-time data acquisition could
which includes algorithms for state of charge and state of be an easy and cheap way to collect field data. These data
health estimation and driving behavior classification. could be used to design the next generation systems in a better
way.
Presently, most of the battery models focus on individual
battery cells, or on battery package not including the detailed VII. CONCLUSION
dynamics within the battery. However, all components in a
In this article we introduced recent advances in industrial
battery are associated with each other. The interactions among
all the cells, conductors, BMS, and environment temperature informatics with respect to Big Data environment, Cyber-
Physical Systems and Industry 4.0. In addition, the importance
will play a significant role in battery performance, and deserve
an integrated study. Moreover, manufacturing practice strongly of intelligent prognostics and health management in industry
for retaining production and service excellence has been
impact on battery application conditions & user behavior.
discussed. Furthermore, the fusion of cyber-physical systems
Therefore a simulated framework is needed which is and PHM algorithms in big data environment and the
capable of execution the functions discussed above by advantage of using interconnected systems has been presented
integrating various models to emulate the multi-regime along with two industrial application of cyber-physical
changes in battery parameters and inputs and investigate their systems.
impacts on battery functionality . Current industrial evolution is guiding industry toward
A "Virtual Battery" is a cyber-physical model resulting maximum leverage from benefits of interconnected systems in
from merging the new battery technology with intelligent big data environment where companies with more futuristic
enabling tools to provide following solutions: vision that establish new methodologies in their culture will
1. Enabling prognostics tools to transform data to health have the opportunity of being significantly successful and
information regarding the health, reliability and operational profitable in recent future.
readiness of the battery system.
 Proceeding of Int. Conference on Industrial Informatics (INDIN) 2014

Cyber-Physical Battery Platform

Historical Intelligent
Data Algorithm

Virtual Real
Condition Condition

Voltage Voltage
Current Current
Temperature Temperature
Speed Speed
GPS data GPS data Predicting Results
Information
EV Weight EV Weight

Simulated Data Raw Data

Driver EV Manufacturer Battery Supplier

SoC Estimation EV Range Material Development
Driving Behavior EV Productivity Battery Safety
Classification EV Safety

Fig. 4. Cyber Physical Battery Platform

and k-nearest neighbor,” 2010 XIX International Conference on Electrical

REFERENCES Machines (ICEM), pp. 1–7, 2010.
[1] L. Da Xu, “Enterprise systems: state-of-the-art and future trends,” [12] E. Lapira, D. Brisset, H. D. Ardakani, and D. Siegel, “Wind turbine
Industrial Informatics, 2011. performance assessment using multi-regime modeling approach,” Renewable
[2] E. Dumbill, “Big Data is Rocket Fuel,” Big Data, 2013. Energy, 2012.
[3] “Adaptive calibration for fusion-based cyber-physical systems,” vol. 11, [13] M. AbuAli, E. R. Lapira, W. Zhao, D. Siegel, M. Rezvani, and J. Lee,
no. 4, pp. 1–25, Dec. 2012. “Systematic Design of Prognostics and Health Management Solutions for
Energy Applications,” in Engineering Asset Management 2011, no. 22, J. Lee,
[4] “Cyber-physical systems,” IEEE Control Systems Society, 2011.
J. Ni, J. Sarangapani, and J. Mathew, Eds. London: Springer London, 2013,
[5] J. Shi, J. Wan, H. Yan, and H. Suo, “A survey of Cyber-Physical pp. 243–250.
Systems,” Signal Processing (WCSP 2011), pp. 1–6, 2011.
[14] B. Bagheri, H. Ahmadi, and R. Labbafi, “Implementing discrete wavelet
[6] “Cyber Physical Systems: the need for new models and design transform and artificial neural networks for acoustic condition monitoring of
paradigms,” Carnegie Mellon University. gearbox,” Elixir Mech Engg, 2011.
[7] J. Lee, E. Lapira, B. Bagheri, and H. Kao, “Recent advances and trends [15] O. D. Snell and I. Nairne, “Acoustic bearing monitoring—the future
in predictive manufacturing systems in big data environment,” Manufacturing RCM 2008,” Railway Condition Monitoring, 2008.
Letters, 2013.
[16] A. Soylemezoglu and S. Jagannathan, “Mahalanobis-Taguchi System as
[8] T. Wang, J. Yu, D. Siegel, and J. Lee, “A similarity-based prognostics a Multi-Sensor Based Decision Making Prognostics Tool for Centrifugal
approach for Remaining Useful Life estimation of engineered systems,” 2008 Pump Failures,” Reliability, IEEE Transactions on, vol.60, no.4, pp.864,878,
International Conference on Prognostics and Health Management (PHM), pp. Dec. 2011.
1–6, 2008.
[17] F. Ju, J. Wang, J. Li, and G. Xiao, “Virtual battery: A battery simulation
[9] E. R. Lapira, “Fault detection in a network of similar machines using framework for electric vehicles,” Automation Science and Engineering, IEEE
clustering approach,” 2012. Transactions on , vol.10, no.1, pp.5,15, Jan. 2013.
[10] L. Liao and J. Lee, “Design of a reconfigurable prognostics platform for [18] Y. He, W. Liu, and B. J. Koch, “Battery algorithm verification and
machine tools,” Expert systems with applications, vol. 37, no. 1, pp. 240–252, development using hardware-in-the-loop testing,” Journal of Power Sources,
2010. 2010.
[11] B. Bagheri, H. Ahmadi, and R. Labbafi, “Application of data mining and
feature extraction on intelligent fault diagnosis by Artificial Neural Network

View publication stats