ACKNOWLEDGEMENT

Through this acknowledgment, I want to express my sincere gratitude to all those people
who gave their views and suggestions for helping in the completion of this project which has
been a learning experience.

At first, I want to express my gratitude to my beloved mentor Prof. Ch. Venkaiah Professor,
Department of Electrical Engineering, for his constant guidance, advices & training which has
played an immense role in the successful completion of this dissertation work. I am very much
indebted to him for suggesting this topic, helping me at every stage and mental support for its
successful completion.

I express my profound thanks to Prof. B. L. Narasimharaju, Professor and Head of the

Department of Electrical Engineering, National Institute of Technology Warangal for providing
necessary facilities and encouragement during the course of the work.

I am also very thankful to the Dissertation Evaluation Committee, for their strenuous efforts
to evaluate my Dissertation. I consider it a great privilege to express my deep gratitude to many
respected personalities who guided, inspired, and helped me in the successful completion of the
Dissertation.

Kuber Kushwaha
Roll no: 22EEM2S04

iv
 ABSTRACT

A accurate model with appropriate parameters is necessary for the design,

simulation, optimization, and evaluation of real-time systems based on proton
exchange membrane fuel cells (PEMFC).

In order to estimate the fuel cell model parameters, this article introduces a
new metaheuristic optimization method called the Polar Light Optimizer (P.L.O.)
and its enhanced version, the Enhanced Polar Light Optimizer (E.P.L.O). The polar
lights' leading properties and the movement patterns observed by celestial the
planets provide the inspiration for the suggested P.L.O. and E.P.L.O. algorithms,
which maximize search efficiency through strategic placement and dynamic
adaptability. Based on these natural events, an algorithmic framework with a
mathematical model is developed for the proposed algorithm.

An objective function based on the difference of the square errors between the
measured and estimated voltage for different data points is developed in order to
choose the best parameters. Two well-known benchmark test suites are used to
determine their parameters using the suggested P.L.O. and E.P.L.O. algorithms.

The statistical results are compared with other well-known algorithms that
have been published in the literature. When statistical values are compared, it is
clear that the suggested P.L.O. and E.P.L.O. algorithms perform better in terms of
accuracy, search capability, and convergence speed .

v
Contents

Acknowledgement iv

Abstract v

List of Figures ix

List of Tables x

Nomenclature xii

1 Introduction 1
1.1 Brief and Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 Organization of the dissertation . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2 Literature Survey 5

3 MODELLING 14
3.1 Overview of the Presented Work . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.2 Flowchart of Operation Strategy . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.3 Time-of-use (TOU) tariff Model . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.3.1 Electric Billing Structure . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.4 Distributed Renewable Energy Sources . . . . . . . . . . . . . . . . . . . . . . 17
3.5 Mathematical Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.5.1 DRESs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.5.2 PEVs Charging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.5.3 Battery Charging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.5.4 Inverter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

4 Optimal Capacity Planning Model 20

4.1 Capacity Planning and Constraints Model . . . . . . . . . . . . . . . . . . . . 20
4.2 Machine Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.2.1 Different Types of Machine Learning . . . . . . . . . . . . . . . . . . 22
4.3 SSML Algorithm Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

vi
 4.3.1 Training Process ........................................................................................ 25
4.3.2 Process to calculate Optimal Solution: ....................................................... 25

5 Case Study 27
5.1 Technical and Economic Data ................................................................................ 27
5.2 Meteorological and Load Consumption Data .......................................................... 27
5.3 Annual Stochastic Data For PEV............................................................................ 29
5.4 Important Input Data Structure ............................................................................... 29

6 Results And Discussion 31

6.1 Analysis of Training Process of SSML ................................................................... 31
6.2 Optimal Capacity With Different Algorithms ......................................................... 31
6.3 System’s components, including Cost of Electricity (COE) and optimized capacity 32
6.4 Convergence Pattern of ML.................................................................................... 33
6.5 Uncertainty Impacts ................................................................................................34
6.6 Advantages of SSML compared to GMDH............................................................. 35

7 Conclusion and Future Scope 37

7.1 Conclusion............................................................................................................. 37
7.2 Future Scope.......................................................................................................... 38

Publications 39

References 40

vii
List of Figures

3.1 Overview of the system grid-connected household. ............................................................ 14

3.2 An HEMS for the proposed model based on real-time rules. .............................................. 15
3.3 Types of Electric billing structure for HEMS. .................................................................... 17

4.1 Visualizing the taxonomy of machine learning: Supervised, Unsupervised, Reinforce-

ment, Semi-Supervised, and Deep Learning. [Internet Source] .......................................... 24

5.1 Real hourly data for 24 hrs wind velocity [7] ..................................................................... 28
5.2 Real hourly data for 24 hrs solar radiation [7]. ................................................................... 28
5.3 Actual power of distributed energy .................................................................................... 28
5.4 Intial SOC of PEV at arrival.............................................................................................. 29
5.5 Crucial input parameters for determining the optimal sizing of PV-battery systems
within grid-connected residential sectors. .......................................................................... 30

6.1 Impact of Training Sample Size for SSML on the COE. .................................................... 32
6.2 Convergence pattern of SSML and GMDH........................................................................ 33
6.3 Optimized COE, SWT capacity, and BSS capacity for the grid-connected home with
fast-charging PEVs are evaluated for uncertainty scenarios 1–10........................................ 34

viii
List of Tables

1.1 Summary of the existing studies on capacity planning . . . . . . . . . . . . . . 3

6.1 Comparison in computing COE Maximum Capacity of Components and time

by various algorithms ............................................................................................. 32
6.2 Components’ Cost of Energy (COE) and optimized capacity incorporated within
the system.............................................................................................................. 33
6.3 Average wind speed, Average Daily Insolation, Average Temperature,and Av-
erage Annual Daily Load in the Scenario for Uncertainty Analysis [18-19]. ............ 35

ix
Nomenclatures

Superscripts

act Actual

chr Charging

dch Discharging

max Maximum limit

min Minimum limit

plt Project lifetime

Subscripts

bss Battery storage system

dpd Dumped power

drs Distributed renewable energy sources

exp Export

hm Home

hmpev Total load, home plus pev

imp Import

ivt Inverters, kW

j Types of components

pev Plug-in electric vehicle

spv Solar photovoltaic, kW

swt Small wind turbine, kW

Abbreviations

AC Annual COE exchange between home and grid ($)

x
GC Capital cost for charger of EV ($)

IOCT Initial operating cell temperature (°C)

TPC Total net present cost of components ($)

Parameters

D Annual degradation (%)

Igs Incident global solar radiation (kWh/m2)

IS Standard insolation (kWh/m2)

Kp Power temperature coefficient (%/°C)

t Time (hr)

Tam Ambient temperature (°C)

Tc Cell temperature (°C)

Yd Derating factor resulting from dust accumulation

β Number of optimized components

E Energy, kWh

P Rated power, kW

SOC State of charge

xi