Content uploaded by Malik Hassan Abdul Rehman

Author content

All content in this area was uploaded by Malik Hassan Abdul Rehman on Oct 25, 2018

Content may be subject to copyright.

1

Demand Side Management Using Hybrid Genetic

Algorithm and Pigeon Inspired Optimization

Techniques

Malik Hassan Abdul Rehman 1, Nadeem Javaid 1,∗, Muhammad Nadeem Iqbal 2, Zaheer Abbas 1,

Muhammad Awais 1, Ahmed Jaffar Khan 1and Umar Qasim 3

1COMSATS Institute of Information Technology, Park Road, Islamabad 44000, Pakistan;

2COMSATS Institute of Information Technology, Wah Cantt 47040, Pakistan;

3Cameron Library University of Alberta Edmonton Canada

E-Mail: ∗nadeemjavaidqau@gmail.com

Abstract—In this paper, our goal is to minimize the

electricity cost, electricity consumption at minimum user

discomfort while considering the peak electricity con-

sumption. Electricity consumption may not be the same

in residential, commercial and industrial areas. It may

vary from each and every area. It is a challenging task to

maintain the balance between the conﬂicting objectives:

electricity consumption and user comfort. To meet the

rising electricity demand in residential area, schedule-

able devices can be equally distributed to the available

time slots on the basis of average power consumption.

The main objective is to minimize the electricity usage

during the electricity peak hours by distributing the

electricity load during the off-peak hours. In this regard,

Genetic Algorithm (GA), Pigeon Inspired Optimization

(PIO) and our proposed hybridization of GA and PIO

(HGP) in Demand Side Management(DSM) are applied

for residential load management to optimize the ﬁtness

function. GA, PIO and HGP are evaluated on the basis of

real time pricing scheme (RTP) for single home with three

different operational time interval (OTI) and for multiple

homes with a single OTI. Simulations results shows that

GA, PIO and HGP are able to minimize electricity bill

and electricity consumption while minimizing the user

discomfort. The performance of HGP is better than GA,

PIO with respect to PAR, electricity load and electricity

cost for both single home and multiple homes scenario.

The feasible region between electricity cost and electricity

consumption is also represented. Moreover, the desired

trade-off between electricity cost and user comfort is also

achieved in both techniques.

Index Terms—demand side management; pigeon in-

spired optimization; genetic algorithm; smart grid;

I. INTRODUCTION

In recent decades, it is being observed that energy

demand is increased immensely. In China and USA,

residential area is considered to be the highest sector

where energy is consumed, as a result huge amount of

greenhouse gas (GHG) is emitted [17]. The traditional

grid has the capability to deliver the electric power

from power generation utility to the consumer. This

one-way communication is not of handling several

parameters of electrical network. In China, residential

area is responsible for most of the power consumption

and greenhouse emissions. In Arabian countries, 40 %

of electricity is consumed in residential area [5]. Bulk

generation and transmission infrastructure is needed

to be installed, to fulﬁll the increase in demand in

residential area. In respond to the high power demand,

electricity price also been increased. To remove the

deﬁciencies of the traditional grid, smart grid (SG) is

emerged to fulﬁll the high energy demand efﬁciently.

Smart grid plays an important role to overcome the

challenges of traditional grid. Smart grid has following

three components: smart meter, smart appliances and

energy management controller. Energy management

is to handle and overcome capacity bounds, environ-

mental issues, maintenance, operations. There are two

categories of energy management: supply side man-

agement (SSM) and demand side management (DSM).

The SSM is responsible for generating, managing and

delivering the energy to the consumers. The DSM is

responsible for managing the energy on the consumer

end. Demand response programs are designed by DSM

that are used in load shifting mechanism in case of

variable prices on different prices on different OTI.

Rasheed et al. in [2] used demand and response (DR)

technique to minimize the cost and PAR while achiev-

ing maximum user comfort. Electricity load shifting

from on-peak to off-peak hours signiﬁcantly reduces

the electricity cost but user comfort is compromised

2

because operation of appliances is delayed [6]. In [11],

electricity load shifting is done by using distributed

algorithm developed by the authors. Residential load

scheduling quandary is proposed utilizing game theory

approach. The convergence rate of Nash equilibrium is

also expedited by the authors by applying the newton

method. Simulation results show that PAR is minimized

while minimizing the user discomfort.

Several aspects such as electricity cost, user com-

fort, electricity load and operational issues has to be

considered in order to achieve coordination between

utility and consumer. User comfort is the important

component that must be considered while reducing the

electricity bill. Lots of efforts has been done in the

literature to handle and overcome these challenges.

In [8], authors used GA to minimize electricity load

in residential, commercial and industrial sector. The

authors done the comparison of the performance of

GA with other Evolutionary algorithms. The results

shows that 21.9 % of electricity load reduced during on-

peak hours. Sahar et al. in [14] presents a new hybrid

approach in which they done hybridization of GA,

BPSO and ant colony optimization (ACO) techniques

for cost minimization, PAR reduction while considering

user comfort on TOU pricing scheme. In [9], authors

proposed a model for handling residential power con-

sumption within user budget. The authors used genetic

algorithm to solve the optimization quandary where

the goal is to increase user comfort while minimizing

electricity cost. Mahmood et al. in [20], presents Real-

istic scheduling algorithm (RSM), that maximizes the

appliance usage at low cost. Simulation results shows

that it maximize appliance usage while minimizing the

electricity cost.

The main goal of this work is to an effective mech-

anism to handle the consumer power consumption and

power demand. We applied GA,PIO and HGP on 14

appliances in a single home and for three different

OTIs: 20 minutes, 30 minutes and 60 minutes. More-

over, we also applied GA, PIO and HGP on 10 homes,

30 homes, 50 homes with 30 minutes OTI and different

power rating. GA,PIO and HGP are evaluated on the

basis of real time pricing scheme (RTP). Simulations

results shows that GA,PIO and HGP are able to mini-

mize electricity bill and electricity consumption while

minimizing the user discomfort. The feasible region

between electricity cost and electricity consumption is

also represented. Moreover, the desired trade-off be-

tween electricity cost and user comfort is also achieve

in both techniques.

II. RE LATE D WO RK

Zhou et al. [1], discussed the contemporary tenden-

cies among HEMS. They present a general overview

on the challenges that occurs while applying HEMS

and the impact of appliance scheduling on the factors

such as, electricity load and user comfort. However, the

impact of integration of RESs and ESSs on electricity

cost is not discussed. Calvillo et al. [3], discussed the

importance of appliance scheduling in HEMS. They

present a general review on the challenges that occurs

during the implementation of HEMS and the impact of

appliance scheduling on PAR. However, user comfort

is not considered in appliance scheduling. Beaudin

and Zareipour [4], discussed the importance of load

scheduling in residential area. They discussed all the

major factors that may effect the electricity cost, PAR,

and electricity load. However they did not considered

user comfort and the impact of user comfort on other

mentioned factors.

In [5], a new hybrid scheme, GAPSO, that is based

on binary particle swarm algorithm(BPSO) and ge-

netic algorithm(GA) is used that is evaluated on day

ahead price for single and multiple days. The pro-

posed hybrid scheme minimizes the electricity bill and

user discomfort. In [6], a new hybrid scheme, TLGO,

that is based on teacher learning based optimization

and genetic algorithm(GA) is used that is evaluated

on day ahead price for single and multiple days.The

proposed hybrid scheme minimizes the electricity cost

at minimum user discomfort or waiting time. The

performance of heuristics techniques is compared with

linear programming (LP) in term of peak electricity

consumption, PAR, electricity bill and user discomfort.

Unlike TLBO and GA, TLGO minimizes both cost

and user discomfort without effecting peak electricity

consumption and PAR. In [7], OHEMS is proposed

that facilitate renewable energy resources and energy

storage system. The results of the proposed scheme and

the heuristic algorithms shows that combining RSS and

RES reduces PAR and electricity bill by 21.55% and

19.94%.

In [8], authors used GA to minimize electricity load

in residential, commercial and industrial sector. The

authors done the comparison of the performance of GA

with other Evolutionary algorithms. The results shows

that 21.9% of electricity load reduced during on-peak

hours. In [9], authors proposed a model for handling

residential power consumption within user budget. To

solve the optimization quandary, the authors used ge-

netic algorithm where the goal is to increase user sat-

isfaction while minimizing electricity bill. Rasheed et

3

TABLE I Strength and limitations of State of the art work

Technique(s) Objective(s) Findings Remarks

GA and BPSO [5] Minimize electricity bill and

user discomfort in single and

multiple homes

Performance of proposed algo-

rithm is compared with evolu-

tionary algorithms

PAR is ignored

TLGO [6] Minimize electricity bill and

user discomfort

Performance of proposed algo-

rithm is compared with LP

Electricity consumption is ig-

nored

OEHMS [7] Minimize electricity bill and

PAR

Performance of proposed algo-

rithm is compared with heuris-

tic techniques

User comfort is not considered

GA [8] Minimize electricity load in

residential, industrial and com-

mercial area

Algorithm is compared with

evolutionary algorithms on the

basis of performance

PAR is effected

GA, BPSO, ACO [14] Minimize electricity cost and

PAR

Algorithm is compared with

evolutionary algorithms on the

basis of performance

User comfort is effected

GA [9] Minimize electricity bill while

maximizing user satisfaction

Manage the load in user de-

ﬁned budget

PAR is ignored

Fractional Programming [10] Minimize electricity bill A novel concept of cost efﬁ-

ciency is proposed

PAR is effected

Distributed algorithm [11] Minimize PAR and user dis-

comfort

PAR is minimized by using dis-

tributed algorithm

Electricity cost is not consid-

ered

MILP [12] Minimize electricity bill and

carbon emmission

Electricity bill is minimized

through MILP

PAR is effected

ILP [13] Minimize electricity cost and

PAR reduction

Electricity cost and PAR are

reduced signiﬁcantly

User comfort and ESS not con-

sidered

al. in [2] used demand and response (DR) technique to

minimize the cost and PAR while achieving maximum

user comfort. Electricity load shifting from on-peak to

off-peak hours signiﬁcantly reduces the electricity cost

but user comfort is compromised because operation of

appliances is delayed [6].

Chen et al. in [10], presented the residential load

scheduling model with RTP. To achieve an optimal

solution for electricity load scheduling, the authors

presented the concept of electricity cost efﬁciency. The

authors used fractional programming for the optimal

solution. Simulation results shows that consumers elec-

tricity cost is reduced. In [11], electricity load shifting

is done by using distributed algorithm developed by

the authors. Residential load scheduling quandary is

proposed utilizing game theory approach. The conver-

gence rate of Nash equilibrium is also expedited by

the authors by applying the newton method. Simulation

results show that PAR is minimized while minimizing

the user discomfort. Mahmood et al. in [20], presents

Realistic scheduling algorithm (RSM), that maximizes

the appliance usage at low cost. Simulation results

shows that it maximize appliance usage while mini-

mizing the electricity cost.

In [12], authors considered 30 homes, where each

home has 12 appliances. The multi-objective optimiza-

tion problem is solved using MILP. Carbon emissions

reduction and consumers electricity bill are achieved.

Shiftable appliances are not considered which plays

a major role in cost minimization. The authors in

[13], proposed an integer linear programming (ILP)

algorithm predicated HEMS with builtin RES to shift

the shiftable electricity load from electricity rush hours

to off-peak hours. User comfort and ESS integration

is not considered by the authors.The authors in [16],

used a decentralized framework to minimize the cost

in residential areas in smart grid by shifting the load

from on-peak to off-peak hours. In [8], authors used GA

to minimize electricity load in residential, commercial

and industrial sector. The authors done the comparison

of the performance of GA with other Evolutionary

algorithms. The results shows that 21.9 % of electricity

4

load reduced during on-peak hours.

III. PROB LE M STATE ME NT

Optimization of energy is the one of the difﬁcult

challenge in smart grid because consumer energy de-

mand and electricity prices are not ﬁxed. In [12],

authors considered 30 homes, where each home has 12

appliances. The multi-objective optimization problem

is solved using MILP. Carbon emissions reduction

and Consumers’ electricity bill are achieved. Shiftable

appliances are not considered which plays a major role

in cost minimization.

In this paper, single home, 10 homes, 30 homes and

50 homes with 14 different appliances are considered.

Three types of appliances are considered: schedule-

able, non-schedule-able and uninterruptable. Our goal

is to minimize the electricity cost, electricity load while

minimizing the user discomfort. We applied GA, PIO,

HGP to achieve our goals. The problem can be stated

as: Given are (a) appliances start and end time (b)

length of operational time (c) RTP signal (d) Time

interval (e) Total power demand of each appliance. To

be determined are (a) power consumption pattern To

ﬁnd the optimal solution with minimum electricity cost,

electricity load and user discomfort, RTP is applied. We

evaluated our model on three different time interval

values: 20 minutes, 30 minutes and 60 minutes. The

four parameter on the basis of which our model is

evaluated are: electricity cost, peak-to-average ration

(PAR), waiting time (user discomfort) and electricity

consumption.

IV. SYSTEM MO DE L

In this paper, a single home with 14 different appli-

ances is considered. Appliances are categorized into

three categories: scheduleable, non-scheduleable and

uninterruptable. Moreover, 20 minutes OTI, 30 minutes

OTI and 60 minutes OTI time slots are considered in

the proposed model. All the appliances have different

length of operation time and energy consumption. All

appliances completes their allocated length of operation

time. Nonscheduleable and uninterruptable appliances

could not be shifted once they start operation. The

proposed system model is shown in Fig. 1.

V. PROPOSED METHODOLOGY

Problem that is stated in section III is solved using

GA,PIO and HGP. In the literature, several mathe-

matical techniques such as LP, MILP and ILP are

used to handle electricity consumption problem. The

computational complexity of mathematical techniques

is very high. We applied population based techniques

to address the electricity consumption problem. We

applied GA, PIO and our proposed HGP techniques

and compared them with previous researchers results.

A. GA

GA is inspired by the genetic process of living organ-

isms. GA has the ability to search for best solution in

minimum time. GA is able to handle complex problems

with minimum computational effort [15]. The initial

process of GA is to generate random population that

updates on every iteration. The status of appliances is

represented by chromosomes and number of hours for

scheduling are represented by length of chromosomes.

Fitness of each chromosome is evaluated based on the

ﬁtness function. The elitism process is performed so

that chromosomes with high ﬁtness value can be used

in next iterations. Two parent chromosomes are selected

after the elitism process is completed. Crossover is

applied to the selected two parent chromosomes and

a child/offspring, that contains the properties of both

the parents, is added to the existing population. In

mutation process the bits of the selected chromosomes

are inverted by mutation operator to reduce the pos-

sibility of repetition of selected chromosomes in the

population. The crossover rate is usually higher than

the mutation rate to get the best possible solution. The

mutation rate is used to maintain randomness to avoid

repetition of same chromosomes. When the crossover

and mutation process is done then ﬁtness of current

population is compared with the previous one until the

termination criteria is achieved. The chromosome with

highest ﬁtness value is selected when the whole process

is terminated.

B. PIO

PIO is derived from homing pigeons and it is pro-

posed by Duan and Qiao [18]. It have two majors

operators: map and compass operator and landmark

operator. Initially the population is randomly generated.

To ﬁnd the best optimal solution, ﬁtness function is

used. All the population is sorted according to the

ﬁtness and half of the population is discarded using

the landmark operator.

C. HGP

Hybridization means to combine two or more tech-

niques [21]. GA,PIO and HGP are combined to form a

hybrid approach, HGP algorithm. All the steps of GA

performed in a same way as discussed earlier but the

5

Fig. 1. System Model

TABLE II Classiﬁcation of Appliances

Groups Appliances Power rating (kWh) LOT

Non Schedule-able appliances Oven 1.3 3 hour

Fan 0.20 15 hour

Kettle 2.0 3 hour

Toaster 0.9 1 hour

Rice Cooker 0.85 2 hour

Blender 0.3 2 hour

Frying Pan 1.1 3 hour

Coffee Maker 0.8 4 hour

Non interrupt-able appliances Washing Machine 0.5 3 hour

Cloth Dryer 1.2 3 hour

Schedule-able appliances Dish Washer 0.7 4 hour

Iron 1.0 3 hour

Vacuum Cleaner 0.4 4 hour

Hair Dryer 1.5 2 hour

elimination and dispersal step of GA in which crossover

and mutation is performed is replaced by the map and

compass operator step of PIO.

VI. SIMULATION RESULTS AND DISCUSSION

In this section we evaluated the performance of GA,

PIO techniques and our proposed hybrid meta-heuristic

technique HGP. We evaluated the following Techniques

on the basis of four performance parameters: Electricity

cost, PAR, Energy consumption and user comfort. We

are considering a single home, consists of 14 appliances

using the RTP price scheme and three different OTI

of 20 minute, 30 minute and 60 minute. We also

considered 10 homes, 30 homes and 50 homes scenario

with 30 minute OTI and different power ratings. The

classiﬁcation of appliances is shown in Table II.

A. Eelectricity Cost

Total cost for single home and multiple homes is

shown in Fig. 2. In single home scenario, GA reduces

4.54%, 2.94%, 20.54% of the total cost in case of

20 minutes OTI, 30 minutes OTI, 60 minutes OTI

repectively. PIO reduces 20.54%, 27.94%, 20.94% of

the total cost in case of 20 minutes OTI, 30 minutes

OTI, 60 minutes OTI repectively. HGP reduces 30.72%,

40.94%, 24.24% of the total cost in case of 20 minutes

OTI, 30 minutes OTI, 60 minutes OTI repectively.

In multiple homes scenario, the cost is less than the

unscheduled cost in all each case.

B. PAR

We applied GA, PIO and HGP on single home with

three different OTIs and on 10 homes, 30 homes and

50 homes with 30 minutes OTI. It is clear from Fig.

3 that our techniques are working ﬁne with respect to

PAR ratio because in all cases the PAR after scheduling

is less than the PAR in unscheduled, it means that our

algorithms are scheduling the appliances and control-

ling the peak so that load is shifted evenly between the

time slots.

6

OTI 20 minutes OTI 30 minutes OTI 60 minutes

0

200

400

600

800

1000

1200

1400

TOTAL COST

Unscheduled

GA

PIO

HGP

(a) Total cost for a single home

10 Homes 30 Homes 50 Homes

0

2

4

6

8

10

Total Cost (Cents)

104

Unschedule

GA

PIO

HGP

(b) Total cost for multiple homes

Fig. 2. Overall cost for single and multiple homes

OTI 20 minutes OTI 30 minutes OTI 60 minutes

0

1

2

3

4

5

6

7

PAR

Unscheduled

GA

PIO

HGP

(a) PAR for a single home

10 Homes 30 Homes 50 Homes

0

1

2

3

4

5

Peak Average Ratio

Unschedule

GA

PIO

HGP

(b) PAR for multiple homes

Fig. 3. PAR for single and multiple homes

OTI 20 minutes OTI 30 minutes OTI 60 minutes

0

1

2

3

4

5

6

7

8

Waiting Time

GA

PIO

HGP

(a) Average waiting time for a single home

10 Homes 30 Homes 50 Homes

0

2

4

6

8

10

12

14

Waiting Time (hours)

GA

PIO

HGP

(b) Average waiting time for multiple homes

Fig. 4. Average waiting time for single and multiple homes

C. User Comfort

We applied GA, PIO and HGP on single home with

three different OTIs and on 10 homes, 30 homes and

50 homes with 30 minutes OTI. It is clear from Fig. 4

that algorithm is working ﬁne with respect to waiting

time because in all cases the waiting time is just the

schedule-able appliances and the non-schedule-able and

uninterruptable appliances does not have any waiting

time it means that user comfort is increased. There is

a trade-off between user comfort and waiting time. In

case of schedule-able appliance the waiting time will

decrease the user comfort but in case of non-schedule-

able and uninterruptable as there is no waiting time

so user will not have to wait, as a result user comfort

increases.

D. Electricity Consumption

We applied GA, PIO and HGP on single home

having 14 appliances and multiple homes scenario. In

single home scenario, we applied these algorithms for

three different OTI to schedule the appliances from

electricity peak hours to off-peak hours. In multiple

homes scenario, we applied these algorithms for 30

minute OTI to schedule the appliances from electricity

peak hours to off-peak hours. It is clear from Fig. 5

that GA, PIO and HGP are able to reduce the peaks

7

0 10 20 30 40 50 60 70 80

Time (hours)

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

Load (kWh)

Unscheduled

GA

PIO

HGP

(a) Single home at 20 minutes OTI

0 5 10 15 20 25 30 35 40 45 50

Time (hours)

0

1

2

3

4

5

6

Load (kWh)

Unscheduled

GA

PIO

HGP

(b) Single home at 30 minutes OTI

0 5 10 15 20 25

Time (hours)

0

2

4

6

8

10

12

Load (kWh)

Unscheduled

GA

PIO

HGP

(c) Single home at 60 minutes OTI

0 5 10 15 20 25 30 35 40 45 50

Time (hours)

0

10

20

30

40

50

60

70

80

90

100

10 Homes Load(kWh)

Unscheduled

GA

PIO

HGP

(d) Power consumption of 10 homes

0 5 10 15 20 25 30 35 40 45 50

Time (hours)

0

50

100

150

200

250

300

30 Homes Load(kWh)

Unscheduled

GA

PIO

HGP

(e) Power consumption of 30 homes

0 5 10 15 20 25 30 35 40 45 50

Time (hours)

0

100

200

300

400

500

600

50 Homes Load(kWh)

Unscheduled

GA

PIO

HGP

(f) Power consumption of 50 homes

Fig. 5. Power consumption of single and multiple homes

8

Algorithm 1 Genetic Algorithm

1: Input: set of appliances Ai;

2: Initialization: P Hs,OP Hs,t=0,

avgga=0, H, PB = 0, 1;

3: for i=1 to T do

4: for j=1 to H do

5: Generate initial population

6: for j=1 to P do

7: Calculate ﬁtness function

8: Select best solution in population P

9: Check status of Ai;

10:

11: if t == P Hs;then

12: wait until OP Hs;

13: Check the remaining t

of all Ai

14: end for

15: Generate new population

16: Crossover (Θi);

17: Mutation (Θi);

18: end for

19: end for

of load. Electricity load peaks have direct impact on

PAR. It is clear from Fig. 5 and Fig. 3, when the peak

electricity load is high then the PAR is high and vice

versa.

E. Performance Trade-off

Performance trade-off is achieved between the elec-

tricity cost and the user discomfort. As shown in Fig. 2

and Fig. 4 when the total cost is high then the waiting

time is low and when the total cost is low then the

waiting time is high. It is clear from Fig. 5 that GA,

PIO and HGP are able to reduce the peaks of load.

Electricity load peaks have direct impact on PAR. It is

clear from Fig. 5 and Fig. 3, when the peak electricity

load is high then the PAR is high and when the peak

electricity load is low then the PAR is low.

F. Feasible Region

Feasible region is a region that contains all the

possible solutions based on our ﬁtness function [19].

Our primary goal is to reduce electricity cost and PAR.

Electricity cost depends on the electricity price and

electricity consumption.We can do load shifting by

shifting the load from on-peak hours where the elec-

tricity price is high to the off-peak hours in which the

electricity price is low.We have to focus on following

four parameters while reducing the electricity cost:

Algorithm 2 PIO Algorithm

1: Input: maximum iterations

2: Initialization: pigeonnum, D, map and compass

factor, T1, T2, Xg

3: Specify LOT of appliances and power ratings

4: Randomly initialized the population

5: set initial path Xiand velocity V for each appliance

6: set Xp=Xi

7: calculate the ﬁtness of individual appliances

8: ﬁnd the optimal solution

9: map and compass operator

10: for i=1:T1 do

11: for i=1:pigeonnum do

12:

13: while Xiis beyond the

search range do

14: calculate Xiand Vi

15: end

16: end

17: for i=1:D do

18:

19: while Xpis beyond the

search range do

20: sort all the appliances

according to their fitness values

21: pigeonnum=pigeonnum/2

22: keep half of the

appliances with better fitness

value and discard the other half

23: Xc= average of the

remaining appliances

24: calculate Xi

25: end

26: end

27: Output: Xgis output as the

global optima of fitness function

28: end

•Minimum Load,minimum price

•Minimum Load,maximum price

•Maximum Load,minimum price

•Maximum Load,minimum price

The blue shaded region in Fig. 6 is the feasible region

for 20 minutes, 30 minute and 60 minute OTI respec-

tively.

VII. CONCLUSION AND FUTURE WORK

In this paper, GA,PIO and our proposed HGP are

applied on single home having 14 appliances and

multiple homes scenario. In single home scenario, we

9

OTI Cases Load (kWh) Price (dollars) Cost (dollars)

Min.load, Min. price 0.462 8.1000 3.7422

Min.load, Max. price 0.462 27.3500 12.6357

20-minutes Max.load, Min. price 3.0167 8.1000 24.4353

Max.load,Max. price 3.0167 27.3500 82.5067

Min.load, Min. price 0.762 8.1000 6.1722

Min.load, Max. price 0.762 27.3500 20.8407

30-minutes Max.load, Min. price 5.925 8.1000 47.9925

Max.load,Max. price 5.925 27.3500 162.0488

Min.load, Min. price 0.462 8.1000 3.7422

Min.load, Max. price 0.462 27.3500 12.6357

60-minutes Max.load, Min. price 10.150 8.1000 82.2150

Max.load,Max. price 10.150 27.3500 277.6025

TABLE III Possible feasible regions for a single home

OTI Cases Load (kWh) Price (dollars) Cost (dollars)

Min.load, Min. price 40.7620 8.1000 330.1722

Min.load, Max. price 40.7620 27.3500 1115

10-homes Max.load, Min. price 500.956 8.1000 4058

Max.load,Max. price 500.956 27.3500 13701

Min.load, Min. price 40.762 8.1000 330.1722

Min.load, Max. price 40.762 27.3500 1115

30-homes Max.load, Min. price 300.0690 8.1000 2431

Max.load,Max. price 300.0690 27.3500 8207

Min.load, Min. price 15.762 8.1000 127.6722

Min.load, Max. price 15.762 27.3500 431.0907

50-homes Max.load, Min. price 101.2070 8.1000 820

Max.load,Max. price 101.2070 27.3500 2768

TABLE IV Possible feasible regions for a multiple homes

Algorithm 3 HGP Algorithm

1: Input: set of appliances Ai;

2: Initialization: P Hs,OP Hs,t=0,

avgga=0, H, PB = 0, 1;

3: for i=1 to T do

4: for j=1 to H do

5: Generate initial population

6: for j=1 to P do

7: Calculate ﬁtness function

8: Select best solution in population P

9: Check status of Ai;

10:

11: if t == P Hs;then

12: wait until OP Hs;

13: Check the remaining t

of all Ai

14: end for

15: Generate new population

16: map and compass operator

17: end for

18: end for

applied these algorithms for three different OTI to

schedule the appliances from electricity peak hours to

off-peak hours. In multiple homes scenario, we applied

these algorithms for 30 minute OTI to schedule the ap-

pliances from electricity peak hours to off-peak hours.

GA,PIO and HGP are evaluated on the basis of real

time pricing scheme (RTP). Simulations results shows

that GA,PIO and HGP are able to minimize electricity

bill and electricity consumption while minimizing the

user discomfort. The feasible region between electricity

cost and electricity consumption is also represented.

Moreover, the desired trade-off between electricity cost

and user comfort is also achieved in existing and

proposed techniques. The performance of HGP is better

than GA, PIO with respect to PAR,electricity load

and electricity cost for both single home and multiple

homes scenario.

REFERENCES

[1] Zhou, B.; Li, W.; Chan, K.W.; Cao, Y.; Kuang, Y.; Liu, X.;

Wang, X. Smart home energy management systems:Concept,

conﬁgurations, and scheduling strategies. Renew. Sustain. En-

ergy Rev. 2016, 61, 30–40.

[2] Rasheed, M.B.; Javaid, N.; Ahmad, A.; Jamil, M.; Qasim, U.;

Alrajeh, N.Energy optimization in smart homes using customer

preference and dynamic pricing. Energies 2016, 9, 125.

10

0123456

Elecetricity Consumption (kWh)

0

50

100

150

200

Electricity Cost ($)

P1(0.70, 7.22)

P2(0.70, 27.86)

P3(1.692, 48.21)

P4(5.89, 7.22)

P5(5.89, 47.9925)

P6(5.89, 162.0488)

(a) Single home 20 minutes OTI

0123456

Elecetricity Consumption (kWh)

0

50

100

150

200

Electricity Cost ($)

P1(0.70, 7.22)

P2(0.70, 27.86)

P3(1.692, 48.21)

P4(5.89, 7.22)

P5(5.89, 47.9925)

P6(5.89, 162.0488)

(b) Single home 30 minutes OTI

0 2 4 6 8 10 12

Power consumption (kWh)

-50

0

50

100

150

200

250

300

Cost per hour ($)

P1(0.7, 2.34)

P2(0.7, 19.65)

P3(2.25, 82.2150)

P4(10.15, 2.34)

P5(10.15, 82.2150)

P6(10.15, 277.6025)

(c) Single home 60 minutes OTI

0 20 40 60 80 100 120

10 Homes Electricity Consumption (kWh)

-500

0

500

1000

1500

2000

2500

3000

Electricity Cost ($)

P1(15.762, 101.207)

P2(15.762, 403.207)

P3(30, 819.7767)

P4(105.207, 819.7767)

P5(105.207, 819.7767)

P6(105.207, 2768)

(d) 10 Homes

0 50 100 150 200 250 300 350

30 Homes Electricity Consumption (kWh)

0

2000

4000

6000

8000

10000

Electricity Cost ($)

P1(40.762, 300.52)

P2(40.762, 1057.22)

P3(87, 2373)

P4(300, 2373)

P5(300, 2373)

P6(300, 8197)

(e) 30 Homes

0 100 200 300 400 500 600

50 Homes Electricity Consumption (kWh)

-2000

0

2000

4000

6000

8000

10000

12000

14000

Electricity Cost ($)

P1(40.762, 247.5)

P2(40.762, 990.62)

P3(147.5, 4008)

P4(5.89, 7.22)

P5(500.956, 4008)

P6(500.956, 13701)

(f) 50 Homes

Fig. 6. Feasible Region for single and multiple homes

[3] Calvillo, C.F.; Sanchez-Miralles, A.; Villar, J. Energy manage-

ment and planning in smart cities. Renew. Sustain.Energy Rev.

2016, 55, 273–287.

[4] Beaudin, M.; Zareipour, H. Home energy management systems:

A review of modelling and complexity.Renew. Sustain. Energy

Rev. 2015, 45, 318–335.

[5] Javaid, N.; Ahmed, F.; Ullah, I.; Abid, S.; Abdul, W.; Alamri,

A.; Almogren, A.S.Towards Cost and Comfort Based Hybrid

Optimization for Residential Load Scheduling in a Smart Grid.

Energies 2017, 10, 1546.

[6] Manzoor, A.; Javaid, N.; Ullah, I.; Abdul, W.; Almogren, A.;

Alamri, A. An intelligent hybrid heuristic scheme for smart

metering based demand side management in smart homes.

Energies 2017, 10, 1258.

[7] Ahmad, A.; Khan, A.; Javaid, N.; Hussain, H.M.; Abdul, W.;

Almogren, A.; Alamri, A.; Niaz, I.A. An Optimized Home En-

ergy Management System with Integrated Renewable Energy

and Storage Resources. Energies 2017, 10, 549.

[8] Bharathi, C.; Rekha, D.; Vijayakumar, V. Genetic Algorithm

Based Demand Side Management for Smart Grid.Wirel. Pers.

Commun.2017, 93, 481–502.

[9] Ogunjuyigbe, A.S.O.; Ayodele, T.R.; Akinola, O.A. User

satisfaction-induced demand side load managementin residen-

tial buildings with user budget constraint. Appl. Energy 2017,

187, 352–366.

[10] Ma, J.; Chen, H.H.; Song, L.; Li, Y. Residential load schedul-

ing in smart grid: A cost efﬁciency perspective.IEEE Trans.

Smart Grid 2016, 7, 771–784.

[11] Li, C.; Yu, X.; Yu, W.; Chen, G.; Wang, J. Efﬁcient computa-

tion for sparse load shifting in demand side management. IEEE

Trans. Smart Grid 2017, 8, 250–261.

[12] Zhang, D.; Evangelisti, S.; Lettieri, P.; Papageorgiou, L.G.

Economic and environmental scheduling of smart homes with

microgrid: DER operation and electrical tasks. Energy Convers.

Manag. 2016, 110, 113–124.

[13] Qayyum, F.A.; Naeem, M.; Khwaja, A.S.; Anpalagan, A.;

Guan, L.; Venkatesh, B. Appliance schedulingoptimization in

smart home networks.IEEE Access 2015, 3, 2176–2190.

[14] Rahim, S.; Javaid, N.; Ahmed, A.; Shahid, A.K.; Zahoor,

A.K.; Nabil, A.; Umar, Q. Exploiting heuristic algorithms to ef-

ﬁciently utilize energy management controllers with renewable

energy sources. Energy Build. 2016, 129, 452–470.

[15] Javaid, N.; Hussain, S.M.; Ullah, I.; Noor, M.A.; Wadood,

A.; Almogren, A.; Alamri, A. Demand Side Management in

Nearly Zero Energy Buildings Using Heuristic Optimizations.

Energies 2017, 10, 1131.

[16] Safdarian, Amir, Mahmud Fotuhi-Firuzabad, and Matti Lehto-

nen. Optimal residential load management in smart grids: A

11

decentralized framework. IEEE Transactions on Smart Grid

7,no. 4 2016, 1836-1845

[17] Geng, Y.; Chen, W.; Liu, Z.; Chiu, A.S.; Han, W.; Liu, Z.;

Zhong, S.; Qian, Y.; You, W.; Cui, X.A bibliometric review:

Energy consumption and greenhouse gas emissions in the

residential sector. J. Clean. Prod. 2017, 159, 301-316.

[18] Yu, M.; Seung, H.H. Supply demand balancing for power

management in smart grid: A Stackelberg game approach.

Applied Energy 2016, 164, 702-710.

[19] Aslam, S.; Iqbal, Z.; Javaid, N.; Aurangzeb, K.; Haider, S.I.

Towards Efﬁcient Energy Management of Smart Buildings

Exploiting Heuristic Optimization with Real Time and Critical

Peak Pricing Schemes. Energies 2017, 10, 1546.

[20] Mahmood, D.; Javaid, N.; Alrajeh, A.; Khan, Z.A.; Qasim,

U.; Ahmed, I.; Ilahi, M. Realistic Scheduling Mechanism for

smart homes. Energies 2016, 9, 1-28.

[21] Longe, O.M.; Ouahada, K.; Rimer, S.; Ferreira H.C.; Vinck,

A.J. Distributed optimisation algorithm for demand side man-

agement in a grid-connected smart micro-grid. Sustain 2017,

9, 998.