Content uploaded by Nadeem Javaid
Author content
All content in this area was uploaded by Nadeem Javaid on Aug 16, 2017
Content may be subject to copyright.
Demand Side Management using DLC in Smart
Grid
By
Fahim Ahmed
CIIT/SP15-REE-010/ISB
MS Thesis
In
Electrical Engineering
COMSATS Institute of Information Technology
Islamabad – Pakistan
Spring, 2017
ii
Demand Side Management using DLC in Smart
Grid
A Thesis Presented to
COMSATS Institute of Information Technology, Islamabad
In partial fulfillment
of the requirement for the degree of
MS (Electrical Engineering)
By
Fahim Ahmed
CIIT/SP15-REE-010/ISB
Spring, 2017
iii
Demand Side Management using DLC in
Smart Grid
A Graduate Thesis submitted to Department of Electrical Engineering as partial
fulfillment of the requirement for the award of Degree of M.S (Electrical
Engineering).
Name
Registration Number
Fahim Ahmed
CIIT/SP15-REE-010/ISB
Supervisor
Dr. Moazzam Islam Tiwana,
Assistant Professor,
Department of Electrical Engineering,
COMSATS Institute of Information Technology (CIIT),
Islamabad Campus.
June, 2017
Co-Supervisor
Dr. Nadeem Javaid,
Associate Professor,
Department of Computer Science,
COMSATS Institute of Information Technology (CIIT),
Islamabad Campus.
June, 2017
iv
Final Approval
This thesis titled
Demand Side Management using DLC in Smart
Grid
By
Fahim Ahmed
CIIT/SP15-REE-010/ISB
has been approved
For the COMSATS Institute of Information Technology, Islamabad
External Examiner: __________________________________________
Dr. Shahzad Saleem
Assistant Professor, Department of Electrical Engineering, FAST National University, Islamabad
Supervisor: ________________________________________________
Dr. Moazzam Islam Tiwana
Assistant Professor, Department of Electrical Engineering, Islamabad
Co-Supervisor: _____________________________________________
Dr. Nadeem Javaid
Associate Professor, Department of Computer Science, Islamabad
HoD: _____________________________________________________
Dr. M. Junaid Mughal
Professor, Department of Electrical Engineering, Islamabad
v
Declaration
I Fahim Ahmed, CIIT/SP15-REE-010/ISB hereby declare that I have produced the
work presented in this thesis, during the scheduled period of study. I also declare that I
have not taken any material from any source except referred to wherever due that
amount of plagiarism is within acceptable range. If a violation of HEC rules on research
has occurred in this thesis, I shall be liable to punishable action under the plagiarism
rules of the HEC.
Date: _________________ Signature of the student:
_____________________
Fahim Ahmed
CIIT/SP15-REE-010/ISB
vi
Certificate
It is certified that Fahim Ahmed, CIIT/SP15-REE-010/ ISB has carried out all the work
related to this thesis under my supervision at the Department of Electrical Engineering,
COMSATS Institute of Information Technology, Islamabad and the work fulfills the
requirement for award of MS degree.
Date: _________________
Supervisor:
__________________________
Dr. Moazzam Islam Tiwana
Assistant Professor
Co- Supervisor:
__________________________
Dr. Nadeem Javaid
Associate Professor
Head of Department:
_____________________
Dr. M. Junaid Mughal
Professor, Department of Electrical Engineering
vii
DEDICATION
I am deeply obliged to acknowledge and thank to those people who put their ever-best
endeavors and contribution in my thesis. First, I am thankful to my Almighty Allah for
blessing me this beautiful life and everything that He has provided me.
Secondly, I cannot forget the appreciation and encouragement from my family
especially from my mother, respected teachers and friends that they gave throughout
my academic life. I also feel great and valuable by being a part of COMSATS
University.
In the end, I humbly extend my special thanks and would like to dedicate this thesis to
my dearest father; Mr. Abdul Shakoor and my beloved son Ahmed Shamim.
viii
ACKNOWLEDGMENT
I am heartily grateful to my supervisor, Dr. Moazzam Islam Tiwana and co-supervisor,
Dr. Nadeem Javaid for their continuous support, motivation, and immense knowledge
from the beginning. Their guidance helped me in doing research and writing of this
thesis.
Besides my supervisor and co-supervisor, I would like to thank comsens research group
who supported me in any respect during the completion of my thesis especially Dr.
Nadeem Javaid for his assistance before, during and at the completion stage of this
thesis work. In the end, I am also thankful to the computer science department for
providing all the facilities during my thesis work.
Fahim Ahmed
CIIT/SP15-REE-010/ISB
ix
ABSTRACT
Demand Side Management using DLC in Smart Grid
In smart grid, several optimization techniques are developed for residential load
scheduling purpose. Most of these conventional techniques of demand side
management aim at minimizing the energy consumption cost. Maintaining a balance
between two conflicting objectives: energy consumption cost and user comfort is still
a challenging task to achieve. Therefore, in this paper, we focus on minimization of
electricity cost and user discomfort while taking into account the peak energy
consumption. In this regard, we implement and analyze the performance of a traditional
technique; dynamic programming (DP) and two heuristic optimization techniques:
genetic algorithm (GA) and binary particle swarm optimization (BPSO) for residential
load. Based on these techniques, we propose a hybrid scheme; GAPSO for residential
load scheduling, so as to optimize the desired objective function. In order to alleviate
the complexity of the problem, the multi-dimensional knapsack is used to formulate the
energy scheduling problem. The proposed model is evaluated based on two pricing
schemes: day-ahead and critical peak pricing for single and multiple days. Furthermore,
feasible regions are calculated and analyzed to develop a relationship between power
consumption, electricity cost, and user discomfort. The simulation results are compared
with DP, and validate that the proposed model along with the proposed hybrid scheme
reflects substantial savings in electricity bills with minimum user discomfort.
Moreover, results also show a phenomenal reduction in peak power consumption.
x
List of Publications
1. Fahim Ahmed, Nadeem Javaid, Ibrar Ullah, Wadood Abdul, Atif Alamri, Ahmed S. Almogren
“Towards Cost and Comfort based Hybrid Optimization for Residential Load Scheduling in
Smart Grid”, 2017 [ Conditionally accepted in Energies’s Journal]. Download
2. Ahmed, Fahim, et al. "Cost and Comfort Based Optimization of Residential Load in
Smart Grid." International Conference on Emerging Internetworking, Data & Web
Technologies. Springer, Cham, 2017. Download
3. Manzoor, Awais, Fahim Ahmed, et al. "User Comfort Oriented Residential Power
Scheduling in Smart Homes." International Conference on Innovative Mobile and
Internet Services in Ubiquitous Computing. Springer, Cham, 2017. Download
4. Judge, Malik Ali, Fahim Ahmed, et al. "Monitoring of Power Transmission Lines
Through Wireless Sensor Networks in Smart Grid." International Conference on
Innovative Mobile and Internet Services in Ubiquitous Computing. Springer, Cham,
2017. Download
5. Fahim Ahmed, Awais Manzoor, Malik Ali Judge.“A review on comfort and energy
management for smart buildigs.” 12th international conference on 2P2, Parallel, Grid,
Cloud and Internet Computing (3PGCIC-2017), Nov 5-7, 2017 in Palau Macaya,
Barcelona, Spain. Download
TABLE OF CONTENTS
List of Acronyms 1
1 Introduction 3
1.1 Introduction............................... 4
1.2 Problem Description . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3 Thesis Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Literature Review 7
2.1 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3 Heuristic Optimization Techniques 13
3.1 Heuristic Optimization Techniques . . . . . . . . . . . . . . . . . . 14
3.1.1 Brief Description of GA . . . . . . . . . . . . . . . . . . . . 14
3.1.2 Brief Description of BPSO . . . . . . . . . . . . . . . . . . . 14
4 Proposed Work 16
4.1 Problem Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.1.1 Multiple Knapsack Problem . . . . . . . . . . . . . . . . . . 17
4.1.2 MKP in Energy Management System . . . . . . . . . . . . . 17
4.2 SystemModel.............................. 18
4.2.1 Energy Management Controller . . . . . . . . . . . . . . . . 19
4.2.2 Communication Network . . . . . . . . . . . . . . . . . . . . 19
4.2.3 Residential Consumers . . . . . . . . . . . . . . . . . . . . . 20
4.3 PricingSchemes ............................ 20
4.3.1 DAPModel........................... 20
4.3.2 CPPModel........................... 22
4.4 Proposed Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.4.1 GAPSO ............................. 22
5 Simulations and Discussion 25
5.1 Simulations and Discussion . . . . . . . . . . . . . . . . . . . . . . . 26
5.1.1 Performance Parameters Definitions . . . . . . . . . . . . . . 26
5.1.2 Peak Power Consumption . . . . . . . . . . . . . . . . . . . 26
5.1.3 Electricity Cost . . . . . . . . . . . . . . . . . . . . . . . . . 27
5.1.4 PAR ............................... 29
5.1.5 UserComfort .......................... 31
5.1.6 Performance Trade-Off . . . . . . . . . . . . . . . . . . . . . 34
5.1.7 Feasible Region . . . . . . . . . . . . . . . . . . . . . . . . . 35
xi
6 Conclusion and Future Works 39
6.1 Conclusion and Future Works . . . . . . . . . . . . . . . . . . . . . 40
7 References 41
xii
LIST OF FIGURES
1.1 Smart Grid Infrastructure . . . . . . . . . . . . . . . . . . . . . . . 4
4.1 An Overview of Home Energy Management System Model . . . . . 20
4.2 PriceSignal ............................... 21
4.3 Electricity Price-CPP . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5.1 Daily Power Consumption-DAP . . . . . . . . . . . . . . . . . . . . 27
5.2 Monthly Power Consumption-DAP . . . . . . . . . . . . . . . . . . 27
5.3 Daily Power Consumption-CPP . . . . . . . . . . . . . . . . . . . . 28
5.4 Monthly Power Consumption-CPP . . . . . . . . . . . . . . . . . . 28
5.5 Daily Electricity Cost-DAP . . . . . . . . . . . . . . . . . . . . . . 30
5.6 Monthly Electricity Cost-DAP . . . . . . . . . . . . . . . . . . . . . 30
5.7 Daily Electricity Cost-CPP . . . . . . . . . . . . . . . . . . . . . . . 31
5.8 Monthly Electricity Cost-CPP . . . . . . . . . . . . . . . . . . . . . 31
5.9 DailyPAR-DAP ............................ 32
5.10MonthlyPAR-DAP........................... 32
5.11DailyPAR-CPP............................. 33
5.12MonthlyPAR-CPP........................... 33
5.13 User Discomfort [Dr:Dryer D.W:Dish Washer W.M:Washing Ma-
chine Ov:Oven Ir:Iron V.C:Vacuum Ket:Kettle To:Toaster R.C:Rice
Cooker H.D: Hair Dryer Bl:Blender F.P:Frying Pan C.M:Coffee
Maker] ................................. 35
5.14 Average Waiting Time [Dr:Dryer D.W:Dish Washer W.M:Washing
Machine Ov:Oven Ir:Iron V.C:Vacuum Ket:Kettle To:Toaster R.C:Rice
Cooker H.D: Hair Dryer Bl:Blender F.P:Frying Pan C.M:Coffee
Maker] ................................. 35
5.15 Feasible Region: Cost and Power Consumption . . . . . . . . . . . 36
5.16 Feasible Region: Cost and Waiting Time . . . . . . . . . . . . . . . 36
xiii
LIST OF TABLES
2.1 Summary of Related Work . . . . . . . . . . . . . . . . . . . . . . . 9
4.1 Abbreviations.............................. 19
4.2 Appliances’ Parameters . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.3 Variables used in Proposed Technique . . . . . . . . . . . . . . . . . 23
5.1 Daily Energy Consumption Cost and Peak Load . . . . . . . . . . . 37
5.2 Performance Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.3 Monthly Energy Consumption Cost and Peak Load . . . . . . . . . 38
xiv
List of Acronyms
SG Smart grid
DSM Demand side management
SSM Supply side management
DR Demand response
RTP Real-time price
ToU Time-of-use
DAP Day-ahead price
IBR Inclined block rate
PAR Peak to average ratio
VOLL Value of load loss
IDR Integrated demand response
LP Linear programming
MILP Mixed integer linear programming
MINLP Mixed integer non-linear programming
GA Geratic algorithm
TLBO Teacher learning based optimization
OSR Optimal stopping rule
BPSO Binary particle swarm optimization
EA Evolutionary algorithm
1
SM Smart meter
EMC Energy Master controller
ESS Energy storage system
RES Renewable energy resources
DER Distributed energy resources
GAMS General algebraic modeling system
PV Photovoltaic
HEMS Home energy management system
HAN Home area network
WAN Wide area network
2
Chapter 1
Introduction
3
1.1 Introduction
The electrical power grid is a gigantic man-made unit. It has often been con-
sidered as the complex engine ever developed. It comprises of synchronous and
asynchronous generators, transformers, transmission lines, relays and switches,
compensators, and controllers etc [1]. There is a dire need to optimize several
components of conventional grid such as unit commitment and generation plan-
ning, economic dispatch, state estimation, load balancing, maintenance scheduling
and dynamic security of the entire electrical network [2].
In this regard, smart grid (SG) is considered to be the most preferable solution to
obviate the aforementioned problems of the traditional grid. The incorporation of
information and communication technologies (ICT) converts CG into SG. The SG
can be referred to as a modern two-way information and power flow system having
the properties of self-healing, resilience and adaptability. The inter-operability for
current and forthcoming standards of devices are ensured in SG that are cyber-
secured against threats [3]. The integration of renewable energy sources (RESs)
and distributed energy resources (DERs) make SG more environment-friendly. SG
provides a mechanism for efficient energy transmission between generation and
consumers by amalgamation of advanced metering infrastructure (AMI), smart
meter (SM), intelligent control system, and advance communication infrastruc-
ture. The SG is responsible for each single activity that is being carried out from
generation unit to end consumers. The infrastructure of smart grid is given in
Fig.1.1
Storage Option
text
text
text
text
text
Smart Grid
Self Healing
Power Quality
Energy Efficiency
Energy Generation
Demand Response
Micro Generations
Large Networks Isolated Microgrids
Power Storage
text
text
Battery Storage
Electricity Market
Figure 1.1: Smart Grid Infrastructure
The world’s pre-eminent fossil resources are at the brink of exhaustion because of
enormous utilization of energy resources. Moreover, around the globe the concerns
over climate change which include ozone layer depletion and global warming are
increasing amongst governments, research communities, policy makers and scien-
tists [4]. In this regard, proper energy management is an essential component to
be addressed. The energy management is classified into two categories: supply
side management (SSM) and demand side management (DSM). The SSM is re-
4
sponsible for generating and delivering reliable energy to the consumers. Whereas,
DSM utilizes the potential of advance communication and control infrastructure.
DSM is one of the key components of the SG that aims at utilizing the available
energy effectively and optimally. DSM designs demand response (DR) programs
which entice the consumers to actively participate in load shifting mechanism in
response to time varying prices. In this way, consumers can achieve significant
cost reduction in their electricity bill [5].
Residential sector has significant contribution in overall energy consumption. More
specifically, this sector is responsible for consuming 22% of US total energy. From
1980-2009, it is observed that residential energy consumption is increased by
57.2%[6]. The unpredicted consumption patterns of consumers in residential sec-
tor may cause the instability of entire network. On the other hand consumers’
convenience may significantly be disturbed. In order to obviate the aforementioned
problems of utility and dwellers, smart homes are introduced which are equipped
with advanced information, control and communication technologies. So as to
optimally utilize the available energy.
Rastegar et al. in [7] proposed an idea of cost minimization along with the value of
lost load (VOLL). The idea behind VOLL is to enhance the consumers’ priorities
and minimize the difference between the actual energy and the predetermined
energy consumption of appliances. In this work, the load is classified into two
classes: controllable and uncontrollable loads. The inclined block rate (IBR) and
time of use (ToU) pricing environment is considered. The mixed integer non linear
programming (MINLP) is used to solve an objective function. In [8], Vardakas et
al. uses the recursive process for peak load calculation. The authors develop four
control scenarios under real-time pricing (RTP) environment.
Authors in [9] and [10] considered energy cost as an objective function to be opti-
mized by efficiently utilizing the available energy. The trade-off analysis between
privacy and cost is addressed in [11], whereas [6] demonstrated a trade-off between
consumers’ comfort and operation delay of devices.
In [12]-[16], residential load scheduling problem is solved by using MINLP. The
primary objective of cost minimization is achieved by using load shifting strategy.
In [17], the authors used game theoretic approach for cost minimization problem,
whereas in [18], a variant of ant colony optimization (ACO) is used to solve the
energy management problem. The optimization techniques work efficiently in
utilization of available energy while respecting all the associated constraints.
In this thesis, a traditional and two heuristic optimization techniques are solved
and implemented for load scheduling problem. Based on these heuristic techniques
a hybrid model is presented with an objective of cost and discomfort minimiza-
tion. Load shifting technique with day-ahead and critical peak pricing (CPP)
mechanism is used. Extensive simulations are conducted to validate the results.
Simulations results validate that our proposed technique performs better in terms
of cost and user discomfort reduction while taking into account the peak demand
reduction.
5
1.2 Problem Description
In literature, zillion of methods are introduced for efficient utilization of available
energy by using DSM infrastructure. The entire electrical network can be made
well balanced and reliable, by managing the energy consumption, electricity cost,
PAR and users’ satisfaction.
In [19] and [20], authors proposed a model for the scheduling of large number
of devices with an objective of cost minimization and reduction of peak power
consumption. Load scheduling strategy is applied in order to achieve an optimal
energy consumption pattern. EA is implemented to apportion the consumers’
load aptly over the time horizon. The proposed models perform well in terms
of cost minimization and peak demand reduction, however, consumers’ comfort
is not addressed, which is a key component for end users’ to participate in DR
programs.
In [21], minimization of electricity consumption cost and user discomfort are con-
sidered as an objective function. Time flexible and power flexible appliances are
considered for efficient utilization of energy. The scheduling problem is formulated
as convex optimization and electricity price is defined by the utility on day ahead
basis. The simulations results show that the proposed technique achieved a de-
sire trade-off between both the parameters of an objective function. However, by
increasing the size of problem computational complexity increases.
In order to minimize the electricity consumption cost along with the maximiza-
tion of consumers’ comfort while taking peak demand reduction into consideration,
there is a dire need in load scheduling mechanism to optimize the aforementioned
parameters. Thus, in this work, we have presented a hybrid technique for the opti-
mal scheduling of residential load and simulations results validate the effectiveness
of the proposed work.
Given are (a) DAP scheme, (b) start and end operation time for devices, (c) task
completion time, (d) power rating of each device, and (e) total time span.
So as to find an efficient energy consumption pattern and an optimal operational
time for devices, with (a) minimum consumption cost, (b) maximum user comfort,
and (c) reduced peak power consumption.
1.3 Thesis Organization
The rest of the thesis is organized as follows. Section 2 contains the related work.
Section 3 presents the heuristic optimization techniques. The proposed work is
discussed in section 4. Section 5 contains simulations and discussions. Section 6
concludes the work.
6
Chapter 2
Literature Review
7
2.1 Literature Review
In SG, DSM is responsible for efficient utilization of available energy. In DSM,
several optimization techniques exist that can efficiently manage the energy con-
sumption behavior of consumers. Many researchers focused at both mathematical
and heuristic optimization techniques which are capable to optimally schedule the
consumers’ load. In [22], the authors proposed an optimization technique: genetic
algorithm (GA), in which cost is taken as an objective function to be minimized.
The results validate that the proposed technique has efficiently and optimally re-
duced the electricity consumption cost of the consumers. Yi et al. in [23] proposed
an opportunistic based optimal stopping rule (OSR) for scheduling of home appli-
ances. Three appliances with RTP are considered in this scheme. This technique
finds an optimal interval where prices are less than a predefined threshold while
waiting time of appliances is taken under consideration. The results show that
consumption cost is reduced significantly with minimum user inconvenience. The
limitation in the proposed technique is the incapability to handle large number of
homes with several devices.
Zhao et al. in [24] presented a model by using GA and used RTP combine with
IBR pricing tariff. The main objective of the proposed technique is to reduce
the electricity cost and peak to average (PAR). In this way, the proposed scheme
strengthened the stability of the entire grid. In residential sector the consumers
have both electrically and thermostatically controllable loads. Various techniques
have been proposed to address this matter. In [25], a scheme: home energy man-
agement for distributed energy resources (DERs) comprising both electrical and
thermal appliances scheduling (HEMDAS) is proposed. The proposed technique
aimed to minimize the energy consumption cost while considering the users’ con-
venience. The MINLP along with dynamic pricing scheme is used. The results
validate that the proposed technique has effectively reduced the cost while consid-
ering the comfort zone of the consumers. In [26], a scheme is proposed in which
residential load is classified into four categories: deferrable, curtailable, thermal
and critical loads. The main goal of the proposed scheme is to minimize the cost
while taking care of comfort zone of the consumers in term of indoor temperature.
The MINLP along with dynamic pricing scheme is used.
The rapid development in DSM shows the emergence of new optimization tech-
niques to create a balance between the power generation and demand in order to
ensure the stability of the grid. In [27], authors proposed optimization techniques:
teacher learning based optimization (TLBO) and shuffled frog leap (SFL). Load
is categorized into three classes: shiftable, sheddable and non sheddable loads.
The proposed scheme aimed at minimizing the electricity cost. In this work three
different pricing schemes are used: ToU, RTP and CPP. The results demonstrated
that the proposed technique has successfully managed to reduce the consumption
cost. In DSM, reduction of electricity cost has remained the primary objective
of the optimization techniques, besides this, waiting time of consumers need to
be addressed. Muralitharan et al. in [28] proposed a multi objective evolution-
ary optimization technique that aimed to minimize the consumption cost while
considering the waiting time of consumers. ToU pricing mechanism is used in
the proposed scheme. The results of the proposed scheme validated the trade-off
8
Table 2.1: Summary of Related Work
Technique(s) Objective(s) Feature(s) Deficiency(ies)
GA [22] Cost minimiza-
tion
Compared the perfor-
mance of GA with two
other algorithms and
through simulations it
has proved that GA
performed well
PAR and con-
sumer comfort
are ignored
OSR with
RTP scheme
[23]
To minimize elec-
tricity cost and
waiting cost
Considered three resi-
dential appliances for
scheduling
Incapable to
deal multi load
scheduling prob-
lem
GA with RTP
and IBR pric-
ing model [24]
To minimize cost
and PAR
Load is categorized
in two classes: man-
ual and auto oper-
ated appliances and
perform the schedul-
ing via heuristic opti-
mization
User comfort is
ignored
MINLP along
with dy-
namic pricing
scheme [25]
To minimize the
cost while consid-
ering user com-
fort
Considered both elec-
trically and thermo-
statically controllable
appliances
PAR is not con-
sidered
MINLP with
dynamic pric-
ing scheme
[26]
To minimize cost
and maximize
thermal comfort
zone
Residential load is
classified as, de-
ferrable, curtailable,
thermal and critical
load
PAR and waiting
time are not con-
sidered
TLBO and
SFL with
ToU, RTP
and CPP
scheme [27]
To minimize the
consumption cost
Load is categorized
as, shiftable, shed-
dble and non shed-
dable load
User comfort
and PAR are
not taken into
consideration
Multi objec-
tive evolution-
ary algorithm
with ToU pric-
ing scheme
[28]
To minimize the
consumption cost
and waiting time
Turn off the devices
when load exceeds
threshold limit and
resume later on
PAR and thermal
comfort level are
not addressed
GA [29] To maximize user
comfort under pre
defined budget
Feasible pattern is
generated so that
consumers acquire
maximum comfort
with minimum cost
Most of the time
appliances having
high power and
cost are com-
pletely ignored
9
Fractional
Programming
with RTP and
DAP scheme
[30]
To minimize the
cost
Introduced the con-
cept of cost efficiency
and analyzed the ef-
fects of DERs and ser-
vice fee on cost effi-
ciency
PAR and user
comfort are not
considered
Multi objec-
tive MINLP
[31]
To minimize the
cost and energy
consumption
The entire system is
improved in term of
thermal comfort
Electrical comfort
and PAR are not
addressed
Quantum
EA is im-
plemented
for optimal
scheduling
and dispatch-
ing of energy
[32]
Optimized the en-
ergy consumption
and integrate the
RESs and ESSs
Efficient utilization of
energy
Electrical and
thermal comfort
are not addressed
MILP and GA
[34]
To minimize the
electricity cost
Incorporate RESs and
ESSs, consumers are
flexible to sale excess
energy to the grid
Efficient energy
scheduling of de-
vices and bidding
among consumers
is ignored
MILP [35] To minimize the
cost and reduce
the carbon emis-
sion
Cluster of 30 smart
homes is considered
each having 12 appli-
ances
User satisfaction
and PAR are ig-
nored
Ordinal po-
tential game
with nash
equilibrium
[36]
To maximize the
consumers incen-
tives and utility
profit
Energy hubs with nat-
ural gas and electricity
are considered
solar and wind
mills are ignored
Game theo-
retic approach
[37]
To minimize the
cost and PAR
Three households are
considered to imple-
ment the proposed
technique
User satisfaction
and RESs are ig-
nored
MILP and
MIQP [38]
To minimize the
cost and modify
the load profile
curve
50 homes are consid-
ered to demonstrate
the model
RESs and DERs
are not addressed
Dynamic
programming
with ToU and
CPP scheme
[39]
To minimize the
electricity cost
The proposed model
has successfully man-
aged to overcome re-
bound peaks forma-
tion problem
User comfort is
not tackled in the
proposed model
between cost and waiting time of consumers. Authors in [29] devised a model that
aimed to maximize the user comfort and minimize the consumption cost. Ogun-
10
PL general-
ized bender
technique [40]
To optimize cost
and occupants’
privacy
large number of appli-
ances are considered
Peak demand re-
duction and oc-
cupants are ad-
dressed
BPSO and
ILP [41]
To minimize the
cost and maxi-
mize the thermal
comfort
Electrically and
thermostatically
controlled loads are
taken
Computational
complexity
juyigbe et al. proposed an optimization technique that is capable to generate an
optimal power consumption pattern, which offered maximum user comfort while
restricting the total cost under the predefined budget. In order to implement this
scheme GA is used due to its flexibility and capability to handle non linearities.
In [30], the authors developed a novel concept of cost efficiency (CE): the ratio of
total energy consumption benefits to the total electricity payments. CE is con-
sidered as an indicator for consumers to adjust their energy consumption pattern.
Moreover, the effects of DERs and service fee on CE are analyzed. The fractional
programming (FP) technique along with day ahead pricing (DAP) and RTP is used
in this scheme. The performance results show that CE is increased with increasing
DERs and decreased with increasing service fee. A multi objective MINLP tech-
nique is developed in [31] to optimize the residential energy consumption pattern.
The main contribution of the Anvari-Moghaddam et al. is demonstrated and val-
idated that the proposed technique successfully managed to reduce the electricity
cost and energy consumption. Moreover, thermal and electrical comfort of the
consumers is taken into consideration. The entire energy management system
is improved in term of thermal comfort because of the introduction of new heat
generation sources and heat flows. Chakraborty et al. [32] devised a system for
energy optimization by the integration of Photovoltaic (PV) and a wind turbine
as RESs. In order to address uncertainties occurred by RESs integration fuzzy
logic is considered. For optimal scheduling and dispatching of energy, an efficient
quantum evolutionary algorithm (EA) is implemented while considering the eco-
nomic and environmental impacts. Moreover, scheduling is performed optimally
in order to alleviate the cost of production and carbon emission resource.
Authors in [33] designed a model to minimize the consumption cost and balance the
energy consumption under ToU pricing scheme. Moreover, renewable and storage
systems are efficiently addressed, so in this way surplus energy can be sold back to
the grid. Three types of costs are considered in this model namely, the installation
cost of the infrastructure, revenue generated from selling electricity to the grid and
cost of purchasing electricity from the grid. Gupta et al. in [34] proposed a model
based on cost minimization problem. MILP is used for problem formulation,
whereas, load scheduling is performed via GA. Additionally, RESs and (ESSs) are
incorporated, as a result of which the load profile of consumers can be modified
and excess energy can be sold back to the grid. In this way, consumers achieved
maximum benefits while utilizing the energy. Zhang et al. in [35] proposed a
model to minimize the electricity cost and reduce carbon emission. A cluster of 30
houses is taken under consideration and each house having 12 devices subjected
to control. Problem formulation is carried out as multi objective optimization.
11
MILP is used to solve this multi objective optimization problem.
A model is proposed in [36], aimed to modify the DR program to integrated DR
(IDR) program. The concept of energy hub is introduced in which different en-
ergy resources are considered. In this work, two energy resources are considered:
electricity from grid and natural gas. Ordinal potential game criteria is used for
coordination among the energy hubs. Consumers are flexible to switch between
the resources at on-peak hours. The proposed model is proven efficient in terms
of maximizing the consumers’ incentives and utility’s profit. Authors in [37] de-
veloped a dynamic model for home energy management system (HEMS). The
proposed model aimed to minimize the consumption cost of residents and to re-
duce PAR for maintaining the stability of a grid. Game theoretic approach is used
for efficient scheduling of residential load. The results validate the performance of
the proposed model. Safdarian et al. in [38] categorized DSM infrastructure into
two stages. In first stage, decentralized system is considered and the aim is to
minimize the electricity cost of consumers. MILP is used to formulate the prob-
lem and is solved by using general algebraic modeling system (GAMS). In second
stage, the aim of the proposed model is to benefice the utility by modifying the
load profile while preserving the constraints of cost and comfort. Mixed integer
quadratic programming (MIQP)is used to achieve the objective of modifying load
profile curve.
Due to the synchronization of consumers’ consumption behavior, there is possibil-
ity of rebound peaks formation at later hours. For optimal utilization of residential
energy a DR program is demonstrated in [39]. Authors proposed a general ob-
jective function, however, considered consumption cost as an objective function
for residential consumers. The results are evaluated and demonstrated that due
to synchronization in load shifting mechanism, unanticipated peaks are created.
The problem is addressed by introducing the concept of multi ToU and multi CPP
mechanisms along with the usage of dynamic programming (DP). In [40], the au-
thors proposed a model based on a large number of residences and appliances.
The proposed model is then formulated to optimize the sum of overall satisfaction
level of consumers in term of cost. PL-Generalized Bender’s technique is used
for scheduling the residential load and protecting the private information of the
consumers. This model efficiently handled the consumption cost of the residential
consumers along with the protection of privacy.
The interval number optimization technique is proposed in [41] to handle residen-
tial load scheduling problem, thermostatically controlled and interruptible loads
are considered in this scheme. Moreover, BPSO combined with integer linear
programming (ILP) is used for load scheduling. This work aimed to minimize
the consumption cost while keeping the thermal comfort under consideration. A
model for home energy management system is proposed in [42]. The main focus of
the work is to minimize the electricity cost and peak energy consumption. While
converting the solar power (i.e., from DC-AC) for households appliances, power
quality is distorted due to harmonics. In order to alleviate the effects of har-
monics, selective harmonic elimination criterion is adopted. The results validate
that substantial savings are achieved in electricity bills while minimizing the peak
power consumption.
12
Chapter 3
Heuristic Optimization Techniques
13
3.1 Heuristic Optimization Techniques
In this section, a brief introduction of heuristic optimization techniques: GA and
BPSO is discussed.
3.1.1 Brief Description of GA
GA aims at finding the best candidates from the entire population. The fittest
candidates are ranked higher in the population, whereas the least fit candidates
are ranked lower in the population. In the end, one fittest candidate is selected
which is called global best. The entire chronological process is followed as, ran-
dom generation of population, fitness evaluation, elitism, selection, crossover and
mutation. The population is updated by using the aforementioned parameters,
the fittest chromosomes are survived and least fit candidates are weeded out in
next population. More detail knowledge about GA can be found in [43, 44].
3.1.2 Brief Description of BPSO
BPSO is a binary variant of (PSO), and is heuristic optimization technique inspired
by the social behavior of bird flocking and fish schooling. BPSO aims at finding
the best possible solution to a problem from entire search space. The velocity
and position of particles are randomly initialized, then updated by using their
respective equations. The particles traverse through the entire space so that an
optimal solution can be found. The evaluation of all the particles are performed
and the global best and personal best positions are updated if required. At the end
of stipulated iterations one global best is opted which is considered as a solution
to the problem [2]. Each particle is associated with its position and velocity. The
position of particle at any point in search space can be determined as,
~
Xk(t) = ~
Xk(t−1) + ~
Vk(t) (3.1)
.
Each particle is associated with the velocity vector, containing the information of
local and global best positions achieved so far. The updated velocity of a particle
can be given as,
~
Vk(t) = ϕ~
Vk(t−1) + Ω1.rand1.(~
Pk−~
Xk(t−1))
+Ω2.rand2.(~
Pg−~
Xk(t−1)) (3.2)
.
where ϕis the inertia constant or weight of the particles, k∈1,2, ..., M is the
number of particles, Ω1and Ω2are constant numbers and Ω1+ Ω2=4. ~
Pkand
~
Pgare local and global best solutions achieved so far. ~
Xk(t−1) and ~
Xk(t) are
previous and current positions of particle in search space.
The velocity update expression composed of three main components.
14
•The first component is often known as “inertia” or “momentum”, it tends
to move a particle in same direction as it was travelling in. The inertia
component can be scaled with a constant factor known as inertia constant.
The inertia constant controls the velocity of a particle so that particle can not
move beyond or below the scope of optimal search space. Mathematically
inertia constant can be given as,
ϕ=ϕf+ (ϕf−ϕi)∗kthiteration
maximum iterations (3.3)
•The second component represents the local best solution found for the first
time in search space. It tends to converge the solution toward local optima.
•The third component can be referred as the linear attraction towards the
global best solution from entire search space. It tends to fetch the optimum
solution by using group knowledge of all the particles.
If the value of velocity exceeds the maximum or minimum limits, then it can be
written as,
~
Vk(t) = ~
Vmax if ~
Vk(t)>~
Vmax;
~
Vmin if ~
Vk(t)<~
Vmin.(3.4)
The position of each member of particle is updated by using the following equation,
~
Xk(t) = 1 if sig(~
Vk(t)> rand);
0 otherwise.(3.5)
where, sig(~
Vk(t)) = 1
(1+exp(~
Vk(t)))
Sigmoid function converts the value of velocity to a binary format by comparing
it with randomly generated number in range between [0 1]. The maximum and
minimum extremes of velocity are [~
Vmax and ~
Vmin].
15
Chapter 4
Proposed Work
16
4.1 Problem Formulation
In this section, energy scheduling problem is formulated for an objective function
and constraints. The aim is to minimize the consumption cost and maximize
the users’ comfort while respecting all the constraints. In objective function we
formulate the maximization of user comfort as minimization of user discomfort,
so both the terms are used interchangeably.
4.1.1 Multiple Knapsack Problem
The energy scheduling is one of the core issues in energy management system. In
this work, multiple knapsack problem (MKP) is used to address the scheduling
problem of a residential load. Knapsack is a combinatorial problem in which a
number of objects, each having weight and value, must be packed in a bin of a
specific capacity, in such a way that the total profit inside the bin is maximum.
MKP is a resource allocation problem, and every resource has a specific capacity
constraint. In this way, the system finds an optimal combination of household
appliances operation modes while respecting the total capacity of available amount
of power [45]. The reasons for using MKP are as follows,
1. It can be referred as a simplest integer linear programming.
2. It can be viewed as subproblems in many complex problems.
3. It may represent the great practical situations.
4.1.2 MKP in Energy Management System
The relationship between key terms used in MKP and energy management system
can be developed as in [46] and is given below,
1. mknapsacks = mtime interval.
2. nobjects = nappliances.
3. wweight of an object = EiEnergy consumed by an appliance i.
4. value of an object = consumption cost of an appliance at time t.
5. Capacity of knapsack = maximum amount of energy that can be drawn from
the grid at time t.
The energy consumed by a single residential device over 24 hour time horizon can
be calculated by using the following equation,
E=
48
X
j=1
Pi
rχij(4.1)
where Pi
ris power rating of a device i and χij is status of device i at time slot j
which can be given as,
17
χij =(1 if device of type i at time j is ON;
0 otherwise.(4.2)
The total energy consumed by the residential devices of all types for N number of
smart homes can be calculated as follows,
Er
t=
48
X
j=1 n
X
i=1
Pi
r×χij !(4.3)
Similarly, the total energy consumption cost of all the devices is given by
Cr
t=
48
X
j=1
n
X
i=1
Pi
r(χij ×λj) (4.4)
The consumers’ discomfort is calculated similar to [21] and is calculated by using
the following equation.
Γ =
n
X
i=1
ρ(T sj
i−T uj
i)k(4.5)
where
0< ρ < 1, k ≥1
PAR can be calculated as follows
P AR =
maxj∈Tn
P
i=1
(Pi
r×χij )
1
48 (
48
P
j=1
n
P
i=1
(Pi
r×χij )
∀T={1,2, ..., 48}(4.6)
The mathematical formulation of an objective function can be given as,
min ω1×Cr
t+ω2×Γ(4.7)
4.2 System Model
The system model comprises of energy management controller, smart homes, com-
munication networks and pricing model. The system model is demonstrated in
Figure 4.1.
18
Table 4.1: Abbreviations
Symbol Description
TTime period of a day
TuUser defined time
TsScheduler defined time
EM C Energy Management Con-
troller
SA Set of residential appli-
ances ∈n
λiElectricity price
lotiLength of operation time
of appliance i
Ti
oti Operation time interval of
appliance i
αiStart time of an appliance
i
βiEnd time of an appliance
CapTDaily energy limit that can
be drawn from grid
4.2.1 Energy Management Controller
In this model, DSM focuses on efficient utilization of energy in residential sector.
The power utility is directly connected to EMC and exchanges bidirectional infor-
mation and unidirectional power flow in real time.The central EMC receives the
price information from the power utility and performs the appropriate action. At
the same time it contains the information from the consumer’s end. It acts as a
gate way between power utility and several homes. The main functionalities of
EMC are monitoring, controlling and managing the residential load. In this case
DSM uses load shifting as a basic scheme that can be implemented by using the
central EMC. In this way EMC is capable to handle large number of residential
appliances in well informed and organized manner. The residential devices send
their arrival requests to the EMC and then requests are processed based on the
availability of time slot. The scheduling mechanism is performed on day ahead
basis.
4.2.2 Communication Network
The communication network includes wide area networks (WANs), neighbourhood
area networks (NANs) and home area networks (HANs). The residential appli-
ances are connected to smart meter via HANs. The residential appliances share
their information to the smart meter and then this information is forwarded to
the central EMC. The smart meters of different homes are connected to the cen-
tral EMC via NANs. Through NANs the collective information is reached to the
main EMC. The EMC exchanges the received information to the power utility
via WANs. Through WANs the demand response and the price information is
19
Appliances
Electrical Power Flow
Communication Network
Power Utility
Smart Meter
Main Pannel
Figure 4.1: An Overview of Home Energy Management System Model
exchanged between power utility and the main EMC.
4.2.3 Residential Consumers
In residential sector, the appliances subjected to control have low energy consump-
tion ratings and short length of operation. There are 2604 controllable appliances
available in this sector from 14 different types of appliances. All types of appli-
ances have different energy consumption pattern and operation time. As in this
area consumers have low priorities regarding the time when the energy has to be
utilized, so more savings can be achieved in residential sector. The amount of
incentives given to consumers depend on how much discomfort the consumer is
willing to undergo. Additionally, in the proposed model comfort level of consumer
is incorporated, as a result of which the consumption cost is increased. Moreover,
half an hour time slot is considered in the proposed model. The power ratings of
appliances and their length of operation are given in Table 4.2.
4.3 Pricing Schemes
In this work two pricing schemes are used, and based on these schemes we analyzed
the performance of our proposed model.
4.3.1 DAP Model
The basic purpose of providing the pricing model on day ahead basis is to facilitate
the consumers to take well-informed decisions. In this way consumers can adjust
20
Table 4.2: Appliances’ Parameters
Appliance’s
Type
Power Rat-
ing (kW)
Length of Op-
eration Time
(hours)
Total De-
vices
Dryer 1.2 4.0 189
Dish Washer 0.7 3.0 288
Washing Ma-
chine
2.0 2.5 268
Oven 1.3 3.0 279
Iron 1.0 2.0 340
Vacuum
Cleaner
2.0 2.0 158
Fan 0.2 24 288
Kettle 2.0 4.0 406
Toaster 0.9 3.0 48
Rice Cooker 0.85 4.0 59
Hair Dryer 1.5 2.0 58
Blender 0.3 1.5 66
Frying Pan 1.1 1.5 101
Coffee Maker 0.8 2.5 56
Total - - 2604
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Time (slots)
0
5
10
15
20
25
30
Price (cents/kWh)
DAP
Figure 4.2: Price Signal
their electricity consumption pattern while taking care of comfort level. This helps
consumers to reduce their electricity bills and these pricing models are readily
available to consumers via advanced metering infrastructure. In this work, DAP
is used similar to [20], and is shown in Figure 4.2. The pricing signal portrays
three main regions: on-peak, off-peak and shoulder-peak hours. The load can be
altered by observing the pricing signal offered by the utility.
21
4.3.2 CPP Model
To validate and generalize the performance of the proposed model, we extend
our approach by implementing the CPP for load scheduling purpose. In CPP,
depending upon the utility policies, electricity prices are double or even higher
at critical peak hours. More specifically in this case we have considered a hot
summer day, having critical peak hours from 12 : 00 pm to 3 : 00 pm where
prices are almost double than usual. Critical events occurred very rarely in entire
season or a year due to intense hot or cold weather. CPP signal is demonstrated
in Fig.4.3.
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Time (Slots)
0
5
10
15
20
25
30
35
40
45
50
Price (cents/kWh)
CPP
Figure 4.3: Electricity Price-CPP
4.4 Proposed Technique
Residential sector has large number of appliances of different types, and all the ap-
pliances have different power ratings and consumption patterns. DSM needs such
a technique that can efficiently handle these complexities. In literature, mathe-
matical techniques such as LP and DP are used for this purpose, these techniques
require more computational time and additionally, inadequate to handle multiple
constraints [20, 47, 48, 49]. Evolutionary heuristic techniques have shown capa-
bilities to cope with such complex scenarios.
4.4.1 GAPSO
In the proposed scheme, the performance of GA and BPSO are evaluated sepa-
rately. The performance of both techniques are analyzed in detail. Finally, com-
bination of these two heuristic optimization techniques is developed and solved.
22
The positive traits of each technique are merged together to avoid the problem of
getting stuck on local optima. The steps involve in the proposed hybrid model
are given as; in the beginning a random population is generated, and then eval-
uation of the fitness function is performed. Tournament base selection criteria is
used for selecting the parents from the population. The crossover and mutation
is done on the selected parents to modify the population. Elitism is the process
of remembering the good solution achieved so far. At this stage, the population
is further updated by using the equations of velocity and position along with
sigmoid function. The updating of the population by using the innate traits of
BPSO further explores the search space. This results in mitigating the problem
of premature convergence. The entire process is repeated until the termination
criteria is reached. The termination criteria depends on the stipulated number of
iterations or when the variations in fitness are not more than a predefined limit
(i.e., 10−10) for numerous (i.e., 50) successive generations.
In this way, the proposed scheme has significantly affected the desired performance
parameters. The user comfort in term of waiting time is also taken into consid-
eration, since it is of great importance. User comfort along with reimbursement
is the only factor which enticed the consumers to actively participate in DR pro-
gram. So, the proposed technique is considered to manage electricity cost and
user comfort along with peak consumption. The parameters used in the proposed
technique are given in Table 4.3.
Table 4.3: Variables used in Proposed Technique
Variables Values
Probability of
crossover
0.9
Probability of
mutation
0.1
Insite 1.0
Vmax 4.0
Vmin −4
k3.0
Ω1 2.0
Ω2 2.0
ρ0.001
Population size 200
Maximum Itera-
tions
600
23
Algorithm 1 GAPSO
begin
Initialize population size, length of chromosome
selection criteria, crossover and mutation rates (pc,pm)
maximum and minimum velocities,maximum number of
iterations, local and global pulls, inertia constant
Generate initial population (X)
Evaluate the fitness function
while(termination criterion not met)
Evaluate the fitness of population
Perform elitism to save the best chromosome
Apply tournament selection criteria to
select two parents from the population
if rand(0,1)≤pc
Select two parents for crossover
Select crossover point of both the parents
Reproduce the offsprings by applying crossover
end if
if rand(0,1)≤pm
Select an individual for mutation
Randomly invert a bit of selected individual
end if
Update Position:
~
Xk(t) = ~
Xk(t−1) + ~
Vk(t)
Update Velocity:
~
Vk(t) = ϕ~
Vk(t−1) + Ω1.rand1.(~
Pk−~
Xk(t−1))
+ Ω2.rand2.(~
Pg−~
Xk(t−1))
Sigmoid Function:
sig(~
Vk(t)) = 1
(1+exp(~
Vk(t))
Evaluate Fitness using equation (4.7)
end while
end
24
Chapter 5
Simulations and Discussion
25
5.1 Simulations and Discussion
In this section, the performance of GA, BPSO, and GAPSO is discussed in de-
tail. While implementing the heuristic optimization techniques for scheduling of
residential load, various factors are observed regarding cost minimization, efficient
power consumption, peak reduction and user comfort.
5.1.1 Performance Parameters Definitions
The cost minimization can be referred as the amount of reduction in electric-
ity bills of consumers. The consumers pay this amount to the utility on hourly
consumption basis at the completion of a predefined period. The efficient power
consumption can be defined as, intelligent utilization of available power in such
a way that the total demand never exceeds the generation capacity. Due to syn-
chronization among consumers’ energy utilization pattern, peaks are formed which
may damage the stability of a grid. The user comfort of consumers can be defined
as, the minimum electricity cost and minimum interruption of devices in daily
routine life.
5.1.2 Peak Power Consumption
Figures. 5.1 and 5.2 show the power consumption behavior under four optimiza-
tion techniques with DAP mechanism on daily and monthly basis respectively.
The energy consumption depends on power rating and length of the operation
time of devices. The performance of GA in term of peak power consumption is
analyzed. It is demonstrated and validated that GA is less efficient when dealing
with peak power consumption. This is due to the global exploration mode of GA
which always focuses on minimum electricity price offered by the utility. This
resulted in peaks formation at off-peak hours while user satisfaction is not taken
under consideration. In this way GA scheduled most of the residential devices at
hours where electricity prices are low regardless of taking peak power consump-
tion into account. Whereas, BPSO performed well in term of reducing peak power
consumption, because BPSO scheduled less number of devices at off-peak hours
as compared to that of GA, this results in significant reduction in peak power
consumption. In GAPSO, the peak power consumption is analyzed and it is ob-
served that peak consumption is reduced to a significant amount. Additionally,
the results of DP are also analyzed, and it is observed that DP performed better in
term of peak demand reduction. It is observed that GA, BPSO, GAPSO and DP
have hourly peak consumption of 1572.3kW , 1232.3kW , 1205.3kW and 1108.8kW
respectively. Results validated that the proposed technique has efficient response
for time varying price signal.
Figures. 5.3 and 5.4 show that the proposed scenario is implemented for CPP. For
CPP, it is observed that during a hot or cold day, most of residents consume energy
during critical peak hours as a result of which more peaks are created during this
time. The over all residential energy consumption behavior is demonstrated in
Table 5.1 and Table 5.3.
26
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Time (slots)
0
200
400
600
800
1000
1200
1400
1600
1800
Power (kW)
Without EMC
GA
BPSO
GAPSO
DP
Figure 5.1: Daily Power Consumption-DAP
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Time (slots)
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
Power (kW)
×104
Without EMC
GA
BPSO
GAPSO
DP
Figure 5.2: Monthly Power Consumption-DAP
5.1.3 Electricity Cost
Electricity consumption cost under different techniques is demonstrated in Figures
. 5.5 and 5.6 for DAP mechanism. It is observed from the figures that the perfor-
mance of GA shows substantial savings in electricity bills. The results validate that
GA achieved 29.9702% reduction in electricity consumption cost. Whereas, BPSO
achieved the reduction of 24.0470% in electricity consumption cost. Because both
27
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Time (slots)
0
200
400
600
800
1000
1200
1400
1600
1800
Power (kW)
Without EMC
GA
BPSO
GAPSO
DP
Figure 5.3: Daily Power Consumption-CPP
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Time (slots)
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
Power (kW)
×104
Without EMC
GA
BPSO
GAPSO
DP
Figure 5.4: Monthly Power Consumption-CPP
the techniques shifted the residential load from on-peak hours to off-peak hours
where prices are minimum regardless of waiting time, and hence results in re-
duction in electricity cost. Through out the ample simulations it is shown that
GAPSO successfully managed to reduce the consumption cost up to 25.2923%
with minimum waiting time. Although the proposed technique is less efficient
than GA in term of cost reduction, however, with optimized consumers’ satis-
faction. The reason associated with this fact is the inverse relationship between
electricity bills and user satisfaction. The performance of the proposed model is
28
also compared with DP. The results demonstrate that the proposed model has
comparable performance with DP, however, with less computational complexity
and storage space.
Since GA finds an optimal or near optimal solution from the entire search space
and schedules the residential devices where consumers pay minimum electricity
expenses. It is an inherent trait of GA that it can deal with complexities and
non-linearities. GA is capable of fulfilling the length of the operation time of all
the devices. Due to all these characteristics GA efficiently manages to reduce the
electricity consumption cost. The performance of BPSO in term of cost minimiza-
tion is analyzed and in this work it is shown that BPSO achieved less savings in
electricity bill as compare to that of GA. It is attributed to the fact that BPSO
scheduled the residential load over the time period uniformly to avoid the peaks
creation. Although BPSO shifted the load at off-peak hours, however, the shifted
load is comparatively less than that of GA. It is worth mentioning that by delaying
an operation of devices, more reduction in electricity cost can be achieved at end
consumers’. While analyzing the performance of the proposed model in term of
cost minimization, it is observed that GAPSO has optimally achieved the objec-
tive of cost minimization. The results show that GAPSO achieved 4.6779% less
reduction in electricity consumption cost as compare to that of GA. Moreover, it
is also observed that GAPSO achieved 1.2453% more reduction in electricity cost
than BPSO, because in proposed technique both the parameters: consumption
cost and user discomfort are taken into consideration. It results in fewer savings
in electricity bills with improved consumers’ lifestyle. To substantiate the perfor-
mance of the proposed work, results are compared with DP. It is observed that
both the techniques performed efficiently while reducing the energy consumption
cost. DP achieved a bit higher savings because it converges to the optimal results,
however, at the expense of time and storage space.
Figures. 5.7 and 5.8 show the energy consumption cost for CPP signal on daily and
monthly basis. It is noted that during critical hours consumers are charged with
high electricity prices. It is also observed that in case of CPP energy consump-
tion cost is significantly increased as compared to that of DAP, because utility
offers maximum electricity prices during critical hours (i.e. 12pm-3pm). Energy
consumption cost is analyzed for both the scenarios in Table 5.1 and Table 5.3.
5.1.4 PAR
The stability and reliability of a grid can be ensured by analyzing the PAR. Fig-
ures. 5.9 and 5.10 depict PAR on daily and monthly basis when considering
DAP as a pricing scheme. Figures infer that GA and BPSO achieved 7.8532%
and 27.7794% reduction in peak power consumption respectively. Both heuristic
techniques scheduled residential load from on-peak hours to off-peak hours. It
is validated from the results that these heuristic techniques scheduled the load
where electricity price is minimum. Whereas, GAPSO reduced 29.3676% peak
power consumption. This is due to the fact that GAPSO managed to distribute
the entire residential load over 24 hours time horizon. The load is distributed
in such a manner that no peaks are created while respecting the waiting time of
29
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Time (slots)
0
50
100
150
200
250
Cost ($)
Without EMC
GA
BPSO
GAPSO
DP
Figure 5.5: Daily Electricity Cost-DAP
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Time (slots)
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
Cost ($)
Without EMC
GA
BPSO
GAPSO
DP
Figure 5.6: Monthly Electricity Cost-DAP
devices. Moreover, the performance of DP is also analyzed and compared with
the proposed approach, it is observed that DP performed better in term of peak
demand reduction.
For CPP, Figures. 5.11 and 5.12 show the PAR for a day and a month respectively.
It is deduced that GAPSO performed better among all the techniques. It is due
to the fact that, GAPSO scheduled the most prior load at critical hours. So as to
maintain the stability of entire electrical network during critical peak hours, while
30
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Time (slots)
0
50
100
150
200
250
Cost ($)
Without EMC
GA
BPSO
GAPSO
DP
Figure 5.7: Daily Electricity Cost-CPP
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Time (slots)
0
2000
4000
6000
8000
10000
12000
Cost ($)
Without EMC
GA
BPSO
GAPSO
DP
Figure 5.8: Monthly Electricity Cost-CPP
taking into account the users’ discomfort.
5.1.5 User Comfort
The user comfort is associated with minimum consumption cost, minimum waiting
time for the operation of devices, maintaining desired indoor temperature level,
illuminance level, air quality and humidity etc. In this work, waiting time is
31
Without EMC GA BPSO GAPSO DP
0
1
2
3
4
5
6
7
PAR
Figure 5.9: Daily PAR-DAP
Without EMC GA BPSO GAPSO DP
0
0.5
1
1.5
2
2.5
3
3.5
4
PAR
Figure 5.10: Monthly PAR-DAP
considered as user comfort and thus to be optimized. While implementing the
GA for the residential load scheduling problem, user comfort is not taken into
consideration. It results in maximum load scheduled at end hours and reduced
maximum consumption cost. Similarly, in BPSO user comfort is not taken into
consideration, and operation time of most of the devices are shifted at later hours.
User comfort in terms of user discomfort and waiting time can be given as,
1. Since in this model, the maximization of user comfort is considered equiva-
32
Without EMC GA BPSO GAPSO DP
0
1
2
3
4
5
6
7
PAR
Figure 5.11: Daily PAR-CPP
Without EMC GA BPSO GAPSO DP
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
PAR
Figure 5.12: Monthly PAR-CPP
lent to the minimization of user discomfort, so both the terms can be used
interchangeably. Fig. 5.13 portrays the user discomfort of all the residential
devices over the 24 hours time horizon. Throughout the extensive simula-
tions results it is noticed that by minimizing the user discomfort, electricity
cost is increased. The waiting time associated with discomfort is also ana-
lyzed and discussed.
2. Fig. 5.14 illustrates that GAPSO successfully managed to reduce the waiting
33
time of appliances. The average waiting time of 5 hours is considered in the
proposed scheme. Moreover in this work, the length of operation time of
fan is 24 hours and it is demonstrated that the associated waiting time
is zero for this device. Generally, by delaying the appliance’s operation
time more monetary benefits are achieved at consumers’ end. It is also
observed in the proposed technique, that with the incorporation of user
comfort, comparatively less savings are achieved. In the proposed scenario
half an hour is considered as an operational time slot of appliances (i.e.,
1slot= 30 minutes).
Fig. 5.13 shows the discomfort faced by each corresponding residential de-
vice. Whereas, Fig. 5.14 shows that average waiting time for each device.
No comparison is being made in these figures, as the purpose of these figures
is to demonstrate the user discomfort and average waiting for each corre-
sponding residential device.
5.1.6 Performance Trade-Off
It is deduced from the results that with the incorporation of user comfort in term
of waiting time, the performance parameters are also affected. It can be viewed
vividly from the same figure (i.e., GA and BPSO) that the user has achieved
maximum monetary benefits, however, compromised on consumers’ convenience.
Similarly, it is shown that GAPSO achieved comparatively less savings in electric-
ity bills with maximum comfort level. In this way, electricity cost and user comfort
both are efficiently addressed in the proposed model. The savings in electricity
bills are decreased by 4.6779%, this decrement in savings is due to the fact that
electricity cost and user comfort are inversely proportional to each other. By in-
creasing the user comfort, savings in electricity bills are decreased and vice versa.
The tradeoff between user comfort and cost is obvious since without sacrificing the
convenience consumers are incapable of achieving the reduction in consumption
cost.
The results show that the proposed technique has tangibly outperformed the exist-
ing scheme. Throughout the simulations, it is observed that the electricity cost and
waiting time along with peak consumption are optimally addressed. The perfor-
mance of the proposed approach is analyzed and compared with other techniques
in Table 5.2. Table shows the upper and lower ranges of energy consumption cost,
user discomfort and peak demand reduction. It is deduced that the proposed
model achieved the desired objective with 95% confidence interval. Moreover, the
optimality of the proposed model is also analyzed, as the DP provides optimal
results. The difference between the performance parameters of proposed model
and that of DP provides the optimality gap. The execution time of the proposed
technique is also analyzed and calculated as 0.4832 seconds which is less than that
of DP having execution time 3.9842 seconds.
34
Dr D.WW.M Ov Ir V.C Fan Ket To R.C H.D Bl F.P C.M
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
User Discomfort
Figure 5.13: User Discomfort [Dr:Dryer D.W:Dish Washer W.M:Washing Machine
Ov:Oven Ir:Iron V.C:Vacuum Ket:Kettle To:Toaster R.C:Rice Cooker H.D: Hair
Dryer Bl:Blender F.P:Frying Pan C.M:Coffee Maker]
Dr D.W W.M Ov Ir V.C Fan Ket To R.C H.D Bl F.P C.M
0
1
2
3
4
5
6
7
8
9
10
Average Waiting Time of Devices (slots)
Figure 5.14: Average Waiting Time [Dr:Dryer D.W:Dish Washer W.M:Washing
Machine Ov:Oven Ir:Iron V.C:Vacuum Ket:Kettle To:Toaster R.C:Rice Cooker
H.D: Hair Dryer Bl:Blender F.P:Frying Pan C.M:Coffee Maker]
5.1.7 Feasible Region
A region comprises a set of points having a possible solution for a problem is known
as a feasible region. Generally, feasible region is associated with the concept of
35
0 200 400 600 800 1000 1200 1400 1600 1800
Power Consumption (kW)
-100
0
100
200
300
400
500
Cost ($)
P1(57.6, 4.6656)
P4(1706.3, 466.67)
P5(1706.3, 207.6448)
P3(1706.3, 138.2103)
P2(57.6, 15.7536)
Figure 5.15: Feasible Region: Cost and Power Consumption
-2 0 2 4 6 8 10
Waiting Time (slots)
-50
0
50
100
150
200
250
Cost ($)
P2(0, 207.6448)
P1(0, 4.6656) P3(10, 4.6656)
P4(10, 97.23)
Figure 5.16: Feasible Region: Cost and Waiting Time
optimization. In this work, feasible region is considered as an area containing all
the possible solutions for an optimization problem. The evaluated performance
parameters are analyzed graphically with the help of feasible region.
Feasible region for consumption cost and power
Electricity cost and power consumption are two directly linked parameters, vary-
ing consumption behavior and electricity price affect the electricity cost. A region
bounded by a set of four points: P1(57.6,4.6656), P2(57.6,15.7536), P3(1706.3,
138.2103) and P4(1706.3,466.67) represents a feasible region for electricity con-
sumption cost and is shown in Fig. 5.15. Point P1(57.6,4.6656) denotes a mini-
36
Table 5.1: Daily Energy Consumption Cost and Peak Load
Technique Parameters Without
EMC
With EMC Reduction
(%)
GA Cost ($) 1581.9 1107.8 29.9702
Peak-
Load(kW)
1706.3 1572.3 7.8532
BPSO Cost ($) 1581.9 1201.5 24.0470
Peak-
Load(kW)
1706.3 1232.3 27.7794
GAPSO Cost ($) 1581.9 1181.8 25.2923
Peak-
Load(kW)
1706.3 1205.3 29.3676
DP Cost ($) 1581.9 1175.6 25.6467
Peak-
Load(kW)
1706.3 1108.8 35.0172
Table 5.2: Performance Analysis
Technique Parameters Lower Value Upper Value
GA Cost ($) 1106.7 1116.0
Discomfort 0.1240 0.8941
PAR 5.8204 6.1599
BPSO Cost ($) 1201.4 1205.2
Discomfort 0.2310 0.8421
PAR 4.5706 4.7336
GAPSO Cost ($) 1179.6 1182.8
Discomfort 0.1102 0.8100
PAR 4.4858 4.5283
DP Cost ($) 1175.6 1175.6
Discomfort 0.1102 0.8100
PAR 4.235 4.235
mum power consumption at minimum electricity cost over the entire day. Whereas,
P2(57.6,15.7536) shows minimum power consumption at maximum electricity cost
offered by the utility. In P3(1706.3, 138.2103), it is demonstrated that the max-
imum consumption at minimum electricity cost. Whereas, P4(17063,466.67) de-
picts an extreme point in a feasible region where both electricity cost and power
consumption are maximum. However, P5(1706.3,207.6448) shows maximum power
consumption and electricity cost for our proposed model. Feasible region infers
that by tailoring the consumption behavior consumers can minimize the consump-
tion cost.
Feasible Region for Cost and Waiting Time
In our proposed scenario, the user discomfort is discussed in term of waiting time
of devices. The maximum allowable waiting time for residential devices is 10 slots
37
(i.e., 5 hours). Fig. 5.16 portrays the trade-off between the consumption cost
and waiting time. User discomfort and electricity cost are inversely proportion to
each other, by decreasing user discomfort electricity cost increases and vice versa.
P1(0, 4.6656) and P2(0,207.6448) show minimum and maximum consumption cost
at zero waiting time. Consumers achieve maximum comfort at zero delay for the
operational time of their devices. Whereas, P3(10,4.6656) and P4(10,97.23) denote
minimum and maximum consumption at maximum waiting time.
Table 5.3: Monthly Energy Consumption Cost and Peak Load
Technique Parameters Without
EMC
With EMC Reduction
(%)
GA Cost ($) 57584 45771 20.5143
Peak-
Load(kW)
41088 32136 21.7873
BPSO Cost ($) 57584 48550.5 15.6883
Peak-
Load(kW)
41088 26928 34.4626
GAPSO Cost ($) 57584 43765 23.9979
Peak-
Load(kW)
41088 27476 33.1288
DP Cost ($) 57584 43840 23.8677
Peak-
Load(kW)
41088 27400 33.331
38
Chapter 6
Conclusion and Future Works
39
6.1 Conclusion and Future Works
In this thesis, we modelled a residential energy management system proposing
a hybrid technique for residential load scheduling. The scheduling problem is
formulated through MKP with a primary objective of minimizing the electricity
cost while maximizing the user comfort. We analysed the performance of our
proposed model under four different parameters: power consumption, electricity
cost, PAR and user discomfort. Furthermore, the performance of the proposed
technique is analysed and compared with GA, BPSO and DP. It is observed that
GA performed efficiently in term of cost minimization whereas, BPSO achieved
significant reduction in peak power consumption. Results demonstrated that the
performance of the proposed model is comparable to that of DP. However, the
proposed model is efficient as it requires less computational time and storage
space. The proportional relation between performance parameters is calculated
and shown with the help of feasible regions. Simulation results show that the
proposed hybrid scheme: GAPSO performed better in terms of cost minimization,
occupants’ comfort maximization along with reduction of peak power consumption
than its counterpart schemes: GA and BPSO.
The security of the smart grid is one of the key factors that most of the residents
are concerned about. So, the proposed technique can be extended to incorporate
security issues. Moreover, renewable energy and storage systems can also be incor-
porated in the proposed model. Last but not the least other energy consumption
sectors can also b included in the proposed model.
40
Chapter 7
References
41
References
[1] Gellings, Clark W. The smart grid: enabling energy efficiency and demand
response. The Fairmont Press, Inc., 2009.
[2] Del Valle, Yamille, et al. “Particle swarm optimization: basic concepts, vari-
ants and applications in power systems.” IEEE Transactions on evolutionary
computation 12.2 (2008): 171-195.
[3] Momoh, James. Smart grid: fundamentals of design and analysis. Vol. 63. John
Wiley & Sons, 2012.
[4] Shaikh, Pervez Hameed, et al. “A review on optimized control systems for
building energy and comfort management of smart sustainable buildings.” Re-
newable and Sustainable Energy Reviews 34 (2014): 409-429.
[5] Gelazanskas, Linas, and Kelum AA Gamage. “Demand side management in
smart grid: A review and proposals for future direction.” Sustainable Cities
and Society 11 (2014): 22-30.
[6] Ahmed, Nadia, Marco Levorato, and Guann-Pyng Li. “Residential Consumer-
Centric Demand Side Management.” IEEE Transactions on Smart Grid (2017).
[7] Rastegar, Mohammad, Mahmud Fotuhi-Firuzabad, and Hamidreza Zareipour.
“Home energy management incorporating operational priority of appliances.”
International Journal of Electrical Power & Energy Systems 74 (2016): 286-
292.
42
[8] Vardakas, John S., Nizar Zorba, and Christos V. Verikoukis. “Power demand
control scenarios for smart grid applications with finite number of appliances.”
Applied Energy 162 (2016): 83-98.
[9] Yalcintas, Melek, William T. Hagen, and Abidin Kaya. “An analysis of load
reduction and load shifting techniques in commercial and industrial buildings
under dynamic electricity pricing schedules.” Energy and Buildings 88 (2015):
15-24.
[10] Zazo, Javier, Santiago Zazo, and Sergio Valcarcel Macua.“Robust Worst-Case
Analysis of Demand-Side Management in Smart Grids.” IEEE Transactions on
Smart Grid 8.2 (2017): 662-673.
[11] Tan, Onur, Jess Gmez-Vilardeb, and Deniz Gndz. “Privacy-cost trade-offs in
demand-side management with storage.” IEEE Transactions on Information
Forensics and Security 12.6 (2017): 1458-1469.
[12] Erdinc, Ozan, et al. “Smart household operation considering bi-directional
EV and ESS utilization by real-time pricing-based DR.” IEEE Transactions
on Smart Grid 6.3 (2015): 1281-1291.
[13] Bradac, Zdenek, Vaclav Kaczmarczyk, and Petr Fiedler. ”Optimal scheduling
of domestic appliances via milp.” Energies 8.1 (2014): 217-232.
[14] Zhang, Di, Nilay Shah, and Lazaros G. Papageorgiou. “Efficient energy con-
sumption and operation management in a smart building with microgrid.”
Energy Conversion and Management 74 (2013): 209-222.
[15] De Angelis, Francesco, et al. “Optimal home energy management under dy-
namic electrical and thermal constraints.” IEEE Transactions on Industrial
Informatics 9.3 (2013): 1518-1527.
[16] Agnetis, Alessandro, et al. “Load scheduling for household energy consump-
tion optimization.” IEEE Transactions on Smart Grid 4.4 (2013): 2364-2373.
[17] Belhaiza, Slim, and Uthman Baroudi. “A game theoretic model for smart
grids demand management.” IEEE Transactions on Smart Grid 6.3 (2015):
1386-1393.
[18] Marzband, Mousa, et al. “Real time experimental implementation of opti-
mum energy management system in standalone microgrid by using multi-layer
ant colony optimization.” International Journal of Electrical Power Energy
Systems 75 (2016): 265-274.
[19] Bharathi, C., D. Rekha, and V. Vijayakumar. “Genetic Algorithm Based De-
mand Side Management for Smart Grid.” Wireless Personal Communications
93.2 (2017): 481-502.
[20] Logenthiran, Thillainathan, Dipti Srinivasan, and Tan Zong Shun. “Demand
side management in smart grid using heuristic optimization.” IEEE transac-
tions on smart grid 3.3 (2012): 1244-1252.
43
[21] Ma, Kai. “Residential power scheduling for demand response in smart grid.”
International Journal of Electrical Power & Energy Systems 78 (2016): 320-
325.
[22] Miao, Hui, Xiaodi Huang, and Guo Chen. “A genetic evolutionary task
scheduling method for energy efficiency in smart homes.” International Re-
view of Electrical Engineering (IREE) 7.5 (2012): 5897-5904.
[23] Yi, Peizhong, et al. “Real-time opportunistic scheduling for residential de-
mand response.” IEEE Transactions on smart grid 4.1 (2013): 227-234.
[24] Zhao, Zhuang, et al. “An optimal power scheduling method for demand re-
sponse in home energy management system.” IEEE Transactions on Smart
Grid 4.3 (2013): 1391-1400.
[25] Shirazi, Elham, and Shahram Jadid. “Optimal residential appliance schedul-
ing under dynamic pricing scheme via HEMDAS.” Energy and Buildings 93
(2015): 40-49.
[26] Althaher, Sereen, Pierluigi Mancarella, and Joseph Mutale. “Automated de-
mand response from home energy management system under dynamic pricing
and power and comfort constraints.” IEEE Transactions on Smart Grid 6.4
(2015): 1874-1883.
[27] Derakhshan, Ghasem, Heidar Ali Shayanfar, and Ahad Kazemi. “The op-
timization of demand response programs in smart grids.” Energy Policy 94
(2016): 295-306.
[28] Muralitharan, K., Rathinasamy Sakthivel, and Yan Shi. “Multiobjective opti-
mization technique for demand side management with load balancing approach
in smart grid.” Neurocomputing 177 (2016): 110-119.
[29] Ogunjuyigbe, A. S. O., T. R. Ayodele, and O. A. Akinola.“User satisfaction-
induced demand side load management in residential buildings with user bud-
get constraint.” Applied Energy 187 (2017): 352-366.
[30] Ma, Jinghuan, et al. “Residential load scheduling in smart grid: A cost effi-
ciency perspective.” IEEE Transactions on Smart Grid 7.2 (2016): 771-784.
[31] Anvari-Moghaddam, Amjad, Hassan Monsef, and Ashkan Rahimi-Kian. “Op-
timal smart home energy management considering energy saving and a com-
fortable lifestyle.” IEEE Transactions on Smart Grid 6.1 (2015): 324-332.
[32] Chakraborty, Shantanu, et al. “Intelligent economic operation of smart-grid
facilitating fuzzy advanced quantum evolutionary method.” IEEE Transactions
on Sustainable Energy 4.4 (2013): 905-916.
[33] Kusakana, Kanzumba. “Energy management of a grid-connected hydrokinetic
system under Time of Use tariff.” Renewable Energy 101 (2017): 1325-1333.
[34] Gupta, Ankita, Bhanu Pratap Singh, and Rajesh Kumar. “Optimal provi-
sion for enhanced consumer satisfaction and energy savings by an intelligent
household energy management system.” Power Systems (ICPS), 2016 IEEE
6th International Conference on. IEEE, 2016.
44
[35] Zhang. “Economic and environmental scheduling of smart homes with micro-
grid: DER operation and electrical tasks.” Energy Conversion and Manage-
ment 110 (2016): 113-124.
[36] Bahrami, Shahab, and Aras Sheikhi. ”From demand response in smart grid
toward integrated demand response in smart energy hub.” IEEE Transactions
on Smart Grid 7.2 (2016): 650-658.
[37] Reka, S. Sofana, and V. Ramesh. “A demand response modeling for resi-
dential consumers in smart grid environment using game theory based energy
scheduling algorithm.” Ain Shams Engineering Journal 7.2 (2016): 835-845.
[38] Safdarian, Amir, Mahmud Fotuhi-Firuzabad, and Matti Lehtonen. “Opti-
mal residential load management in smart grids: A decentralized framework.”
IEEE Transactions on Smart Grid 7.4 (2016): 1836-1845.
[39] Muratori, Matteo, and Giorgio Rizzoni. “Residential demand response: dy-
namic energy management and time-varying electricity pricing.” IEEE Trans-
actions on Power systems 31.2 (2016): 1108-1117.
[40] Moon, Seokjae, and Jang-Won Lee. “Multi-Residential Demand Response
Scheduling with Multi-Class Appliances in Smart Grid.” IEEE Transactions
on Smart Grid (2016).
[41] Wang, Jidong, Yinqi Li, and Yue Zhou. “Interval number optimization for
household load scheduling with uncertainty.” Energy and Buildings 130 (2016):
613-624.
[42] Rajalingam, S., and V. Malathi. “HEM algorithm based smart controller for
home power management system.” Energy and Buildings 131 (2016): 184-192.
[43] Kim, D.H., Abraham, A. and Cho, J.H., 2007. A hybrid genetic algorithm
and bacterial foraging approach for global optimization. Information Sciences,
177(18), pp.3918-3937.
[44] Arabali, A., Ghofrani, M., Etezadi-Amoli, M., Fadali, M.S. and Baghzouz, Y.,
2013. Genetic-algorithm-based optimization approach for energy management.
IEEE Transactions on Power Delivery, 28(1), pp.162-170.
[45] Morimoto, Naoyuki. ”Energy-on-Demand System Based on Combinatorial
Optimization of Appliance Power Consumptions.” Journal of Information Pro-
cessing 25 (2017): 268-276.
[46] Kumaraguruparan, N., H. Sivaramakrishnan, and Sachin S. Sapatnekar.
“Residential task scheduling under dynamic pricing using the multiple knap-
sack method.” Innovative Smart Grid Technologies (ISGT), 2012 IEEE PES.
IEEE, 2012.
[47] Ng, K.H. and Sheble, G.B., 1998. Direct load control-A profit-based load
management using linear programming. IEEE Transactions on Power Systems,
13(2), pp.688-694.
45
[48] Kurucz, C.N., Brandt, D. and Sim, S., 1996. A linear programming model for
reducing system peak through customer load control programs. IEEE Trans-
actions on Power Systems, 11(4), pp.1817-1824.
[49] Hsu, Y.Y. and Su, C.C., 1991. Dispatch of direct load control using dynamic
programming. IEEE Transactions on Power Systems, 6(3), pp.1056-1061.
[50] Azadeh, A., Ghaderi, S.F., Tarverdian, S. and Saberi, M., 2007. Integration
of artificial neural networks and genetic algorithm to predict electrical energy
consumption. Applied Mathematics and Computation, 186(2), pp.1731-1741.
46
