electronics

Article
Optimizing Energy Consumption in the Home Energy
Management System via a Bio-Inspired Dragonfly
Algorithm and the Genetic Algorithm
Irshad Hussain 1 , Majid Ullah 1 , Ibrar Ullah 1 , Asima Bibi 1 , Muhammad Naeem 1 ,
Madhusudan Singh 2, * and Dhananjay Singh 3, *
1 Faculty of Electrical & Computer Engineering, University of Engineering and Technology, Peshawar 25000,
Pakistan; [email protected] (I.H.); [email protected] (M.U.);
[email protected] (I.U.); [email protected] (A.B.);
[email protected] (M.N.)
2 Endicott College of International Studies, Woosong University, Daejeon 300-718, Korea
3 Department of Electronics Engineering, Hankuk University of Foreign Studies, Yongin 17035, Korea
* Correspondence: [email protected] (M.S.); [email protected] (D.S.)




Received: 13 January 2020; Accepted: 15 February 2020; Published: 28 February 2020


Abstract: Due to the exponential increase in the human population of this bio-sphere, energy resources
are becoming scarce. Because of the traditional methods, most of the generated energy is wasted every
year in the distribution network and demand side. Therefore, researchers all over the world have taken
a keen interest in this issue and finally introduced the concept of the smart grid. Smart grid is an ultimate
solution to all of the energy related problems of today’s modern world. In this paper, we have proposed
a meta-heuristic optimization technique called the dragonfly algorithm (DA). The proposed algorithm
is to a real-world problem of single and multiple smart homes. In our system model, two classes of
appliances are considered; Shiftable appliances and Non-shiftable appliances. Shiftable appliances
play a significant role in demand side load management because they can be scheduled according to
real time pricing (RTP) signal from utility, while non-shiftable appliances are not much important in
load management, as these appliances are fixed and cannot be scheduled according to RTP. On behalf
of our simulation results, it can be concluded that our proposed algorithm DA has achieved minimum
electricity cost with a tolerable waiting time. There is a trade-off between electricity cost and waiting
time because, with a decrease in electricity cost, waiting time increases and vice versa. This trade-off
is also obtained by our proposed algorithm DA. The stability of the grid is also maintained by our
proposed algorithm DA because stability of the grid depends on peak-to-average ratio (PAR), while
PAR is reduced by DA in comparison with an unscheduled case.

Keywords: optimization; demand side management; demand response; dragonfly algorithm; energy
management controller; energy management system; genetic algorithm; smart meter; smart grid;
traditional grid; peak to average ratio

1. Introduction
In today’s developing era, everything is changing very quickly. In all aspects of life, things are
changing and improving to provide maximum comfort to the end users. Every researcher is exploring
the unseen world to use it for benefits of human beings. In those research fields, one major research
area is energy optimization. Electrical energy is used for so many purposes such as in our homes,
offices, industries, commercial buildings, educational institutes, research labs, vehicles, hospitals,
and even in transportation nowadays. It is quite noticeable that the electricity usage is increasing day
by day. Every day brings more buildings, such as power consumption centers, and the amount of

Electronics 2020, 9, 406; doi:10.3390/electronics9030406 www.mdpi.com/journal/electronics

Electronics 2020, 9, 406 2 of 17

available energy is constant because a power station produces only a specified amount of energy for
a specified period of time and the amount of generating energy can only be increased if the structure of
the entire generating station is changed, which is impossible. The whole structure can not be changed
due to intensification in energy demand. To overcome this problem, different techniques have been
used until now. These techniques or algorithms are of a different nature; some of them are based
on mathematical models, some are nature inspired algorithms, and some of them are bio-inspired
algorithms. Researchers model their problem according to these algorithms and try to find ideal or
near ideal solutions to their problems.
As mentioned earlier, a limited quantity of energy is usually available for most of the users.
That is why we need to utilize the energy intelligently and efficiently. As a result, the idea of Smart
Grid (SG) was presented. SG is not much more than a Traditional Grid (TG), but, before interpreting
an TG as an SG, certain additional technologies must be integrated into TG like Smart/Intelligent
Control Systems, Information Technology, and Cyber security, which, in turn, make a Cyber-Physical
System as exploited by a Smart Grid Network [1], which is mandatory for the reliable operation of
a SG. That’s why the term ‘smart’ is used for it. SG is actually a complicated structure composed
of many sections. A service is offered by SG called Demand Side Management (DSM) where the
appliances operate according to the RTP signal. DSM is the modification of consumer’s demand for
energy consumption. DSM provides two services ‘Energy Management’ and ‘Demand Response’ (DR).
DR is a financial strategy used by electric utility companies to compel their users to reduce or shift
energy consumption from high cost hours to low cost hours of the day. DSM and DR focus on the
intelligent utilization of energy resources. DR provides different benefits to the end electricity users [2].
It can be achieved through two techniques: Incentive-based and Price-based [3]. In an incentive-based
technique, the user’s appliance is switched to an ON/OFF state by sending a short message to the
smart home (SH); when a high peak is detected, the appliance is switched to off state and vice versa.
Thus, in this way, the PAR is decreased by the company, which is the main objective in an optimization
problem. In the price-based program, the user is motivated to use their appliances during low price
hours or off-peak hours. If the user schedules his appliances in low cost hours; as a result, he will
be charged less for one unit of electricity which will bring the effective decrease in his electricity bill.
Actually, a trade-off among cost and user waiting time exists, so scheduling can disturb the user’s
comfortability if the scheduling is not intelligent. Thus, a user can decrease his energy consumption
and PAR via cooperative relationship between the SM and the electric company [4]. According
to [5], electricity consumption can be decreased 10–30% by scheduling of appliances intelligently.
It shows that scheduling can perform well in the objectives of an optimization problem. SG also
integrates renewable energy resources (RES) and sensors to make the procedures more user friendly
and transparent [6]. RES are used in emergencies and in situations whenever catastrophes occur.
Researchers try to capture the natural phenomenon in their algorithms, by capturing several
nature-inspired, natural phenomena, bio-inspired and meta heuristic inspired algorithms were
developed. They have also explored and analyzed some other algorithms for problems related
to energy optimization like: Bio-inspired Genetic Algorithms (GAs) and Cuckoo Search Algorithm
(CSAs) [7]. Therefore, a new meta-heuristic optimization algorithm is proposed in this research paper,
named DA [8], for minimization of consumed energy cost, PAR, and user waiting time. The simulation
results are compared with a well-known optimization algorithm, GA.

2. Literature Review
In the past few decades, particularly in the last two decades, a lot of work has been carried
out in the research domain of energy optimization. Different algorithms have been proposed and
implemented successfully in the field of energy optimization. In [9], the authors have proposed
an hour-ahead DR algorithm for EMSs in a home. In this work, the authors have presented a steady
price prediction model, whose working principle is based on ANN, that is used for dealing with the
uncertainty in pricing signals from utility in the future. The home load is categorized as shiftable,
Electronics 2020, 9, 406 3 of 17

non-shiftable, and controllable appliances. They have reduced the user electricity bill and discomfort
due to the scheduling of appliances. However, PAR is not taken into consideration. In [10], the authors
have used TLBO, EDE, and their hybrid version EDTLA for an efficient EMS to deal with energy cost
and minimize user discomfort in terms of waiting time. In [11], an HEMS is proposed using EDE,
HSA and their hybrid v for a version named harmony EDE (HEDE) for optimal use of the existing
resources of energy to reduce cost and PAR. However, they did not include the RES to save energy.
Wu et al. [12] have proposed a stochastic dynamic programming framework for efficient HEMS with
PEV energy storage. They have reduced electricity cost. However, PAR is not taken into consideration.
Similarly, in [13], the authors have used GA, MFO, and hybridization of these two, named TG-MFO
for efficient EMS in homes. In [14], the authors combined wind, PV, and fuel cells to make a hybrid
energy system, by using PSO. The main theme of this work was to achieve maximum energy using
a controlling strategy. In [15], the authors have used WDO, HSA, GA, and hybridization of GA and
HSA, named GHSA for efficient HEMS to reduce the electricity bill and PAR. In [16], the authors have
proposed a hybrid version of GA and TLBO, named TLGO, to achieve minimum energy cost and PAR.
In [17], the authors have used GOA and CSA for efficient EMS in industries to reduce cost and PAR.
In [18], the authors have categorized the home appliances as power flexible and time flexible in
order to achieve maximum reduction in the electricity cost and reduce the waiting time. In [19],
the authors have proposed a hybrid optimization algorithm, named bird-mating optimization,
for achieving voltage balancing and mitigation of the deviation, so that power could be supplied
efficiently. In [20], the authors have proposed a hybrid version of CSA and EWA. They have applied
the concept of re-scheduling the appliances, if they are interrupted after scheduling, due to which
the authors have claimed 50.6% reduction in the cost, using RTP signaling. In [21], the authors have
depicted a review of the research articles using heuristic techniques for energy optimization in smart
homes. In [22], the authors have proposed GOA and BFA for EMS in an office, for reduction of
electricity bill and PAR. The authors have claimed 34.69% reduction in electricity bill using GOA and
37.47% reduction in electricity bill using BFA. Similarly, in [23], the authors have proposed GA with
RTP and IBR price signaling for reduction of electricity costs and PAR. Furthermore, solar energy
as RES is also added for further reduction of electricity cost. In [24], the authors have analyzed and
compared GA and CSA for total cost, PAR reduction, and waiting time, where he simulated the
results through MATLAB and presented that the cost is reduced by 22.84% using GA and 21.47%
reduction is achieved using CSA, while PAR is reduced to 3.63 for GA (18.24% reduction) and 3.7198
for CSA (19.00% reduction). From these results, it is clear that GA is better for costs while CSA is
better for reduction of PAR. In all of this research, the algorithms carry out scheduling of appliances in
accordance with the electricity price signal provided by the electric utility company. Every electricity
price signal comprises high and low peak hours designed by the utility. The electricity price signal
may be of different time slots; in some works, the authors considered electricity price signals of 48
time slots and some authors considered the electricity price signal of 24 time slots. The electric utility
companies encourage the costumers to operate their appliances in low peak hours instead of random
operation or operating in high peak hours. In this way, the end users or customers are charged less as
compared to un-scheduled load cost. Therefore, all algorithms try to shift the appliances to off-peak
hours from high peak hours of the day. In [25], the authors have proposed CP DR energy optimization
in a smart home for energy cost reduction. They have transformed MIP to CP for efficient optimization
of household appliances. In [26], the authors have proposed an optimal scheduling for micro-grid DG
units, using MVPA to reduce operational cost. In [27], the authors have proposed a modified version
of CSA, i.e., a self-adaptive step-size version for solving ED issues, particularly, for large-scale systems.
In [28], the authors have proposed a GWO algorithm for efficient energy optimization, and claim
33.185% reduction in the operational cost. In [29], the authors have proposed a microgrid connected
with main grid for reducing energy cost and PAR using GWO, BPSO, GA, and WDO optimization
techniques. They have also made hybrid versions of different algorithms for getting more good results.
In [30], the authors have proposed a DC micro-grid, which consists of PV, fuel cells, and batteries for
Electronics 2020, 9, 406 4 of 17

smart homes EMS to reduce the generation cost. They have used ABC for this purpose. In [31], the
authors have used GA and BPSO for efficient HEMS. The authors have proposed a hybrid version
named GAPSO to improve their results. This literature review gives an idea behind the use of different
optimization algorithms for energy cost and PAR reduction intelligently and efficiently. However,
there is always a trade-off between cost reduction and consumer discomfort in terms of waiting time,
when appliances are scheduled from on peak hours to off peak hours.

3. Proposed System Model

Efficient utilization of energy in SG is authorized by DSM. DR and DS Control activities are
the key factors of DSM for customers. In this research, a system model is proposed for a single
utility and number of users. For computing hourly power consumption of every end user, SMs
are installed in every SH. The purpose of the smart meter is to communicate the power consumed
by the user to the utility, communicate the pricing signal to the EMC, and also to communicate
the feedback of each appliances to the utility for the billing purpose and other benefits of the
consumer and utility. The power utility is connected to the EMC. In EMC, there is bidirectional
communication flow and unidirectional power flow in real time. We have considered a set of
12 appliances A App = A App1 , A App2 , A App3 , ..., A Appn , in our proposed system model. It is assumed
that every appliance of our system model is capable of communicating with EMC using GSM
technology. The purpose of EMC is to schedule all the appliances according to the pricing signal
provided by the utility. The complete details of the system model are shown in Figure 1.

Figure 1. Proposed system model.

For the goal of optimization to be achieved, each appliance should complete its operation in the
scheduled time slot. Here, in this work, we considered 24 time slots per day, one time slot is equal
to 1 h, and the unit price of each hour is varied according to the RTP signal. Furthermore, we have
classified our appliances to two classes, which is explained in the next section.

3.1. Classification of Load

For evaluation of the objective function of our suggested scheme, we assumed two real-world
scenarios in our research; first, we check the performance of our suggested scheme for single and
multiple smart homes. In each smart home, we assumed a set of appliances; for simplicity, we have
classified our appliances into two classes (Class A and Class B). In Class A, we have placed shiftable
appliances while, in Class B, we have placed non-shiftable appliances. For every smart home,
Electronics 2020, 9, 406 5 of 17

let An = as + ans strates two classes of appliances, where as represents shiftable appliances, and ans
represents non-shiftable appliances, respectively. The time horizon of one day is explained below:

teT, ∀, T = t1, t2, t3, t4, ...., t24.

Normally, in every home, people use different appliances according to their daily needs. In every
home, the living style and habits of people vary according to their routine, which is why the energy
usage and length of operational time (LOT) of their appliances must be different from one another.
In this research work, we assumed 12 different appliances in every smart home. The power rating,
starting time, finishing time, and LOT of each appliance are depicted in Table 1.

Table 1. Appliances categorization, names, power rating, starting time, finishing time, and Length of
operational time (LOT).

Appliances Power Rating Starting Time Finishing

Appliance Name LOT (h)
Class (kW) (h) Time (h)
Coffee maker 1.0 08 10 1
Printer 0.5 18 20 1
Microwave oven 1.7 08 10 1
Laptop 0.1 18 24 2
Shiftable Desktop 0.3 18 24 3
Appliances Vacuum Cleaner 1.2 09 17 1
Electric Car 3.5 18 08 3
Iron 0.8 09 17 2
Washing Machine 1.5 09 12 2
Hair Dryer 1.5 13 18 1
Non-Shiftable Interior Lightening 0.84 18 24 6
Appliances Refrigerator 0.3 8 8 24

3.1.1. Shiftable Appliances

Shiftable appliances is a class of those appliances, that can be shifted to any time slot in 24 h.
In addition, it is also possible to interrupt the operation of these appliances. We placed the coffee maker,
printer, microwave oven, laptop, desktop, vacuum cleaner, electric car, iron, washing machine, and the
hair dryer in this group. Let as be the combination of shiftable appliances and As eas demonstrates
all the appliances in this class as shiftable appliances. Here, λs is the power consumption of each
appliances in this class ρ(t), presents the unit price, es , which represents a time horizon of one
day, and α a (t) shows the Active/Inactive status of the appliances. The total energy consumption
and its respective cost for one day for a single home, and thirty homes is shown in the following
mathematical relations:
Total energy consumption:
24
Es = ∑ αs eAn( ∑ λs × ×αs (t)) (1)
t =1

Total cost of consumed energy for shiftable appliances in a single home:

24
σα Totals = ∑ αs eAn( ∑ λs × ρ(t) × αs (t)) (2)
t =1

Total cost of consumed energy for shiftable appliances in thirty homes:

30 24
ϕα Totals = ∑ (∑ αs eAn( ∑ λs × ρ(t) × αs (t)) (3)
Home=1 t =1
Electronics 2020, 9, 406 6 of 17

where α(t) represents the ON/OFF state of the appliance and ρ(t) represents the unit price.

3.1.2. Non-Shiftable Appliances

This class of appliances are also called fixed load appliances. It means that, shifting from one time
slot to other time slot or interrupting during operation time is not possible in such type of appliances.
Non-shiftable appliances need a fixed time for the finishing of their operational time slot. Refrigerator
and interior lightening are placed in this class. Let αns show each of the nonshiftable appliances in
Ans set of non-shiftable appliances. Here, λns represents the power consumption of each non-shiftable
appliance, ρ(t) represents the unit price, ens presents a time horizon of one day, and αns (t) represents
the Active/Inactive status of appliances in this class. Total energy consumption of one day is shown in
the following mathematical relation:
Total Energy Consumption:
24
ens = ∑ αns eAn s( ∑ λns × αns (t)) (4)
t =1

An end user executes these appliances in an un-scheduled manner, due to the non-shiftable nature
of appliances in this class, the customers pay a high amount because the demanded time slot is high
pricing time. A high price is paid because of high PAR value. To retain the equilibrium between
energy generation and consumption, the utility charges high prices for consumption of electricity
in demanded hours. For non-shiftable appliances, the cost consumption for one day of single and
multiples homes is shown in the following mathematical relations:
Cost Consumption of single home:

24
σα Totalns = ∑ αns eAn( ∑ λns × ρ(t) × αns (t)) (5)
t =1

Cost Consumption of thirty homes:

30 24
ϕα Totalns = ∑ ( ∑ ( ∑ λns × αns (t))) (6)
Home=1 αnseAn t=1

ϕα Total = ϕα Totals + ϕα Totalns (7)

4. Pricing Signal
Many international energy system operators issue hourly Real Time Pricing (RTP) signals every
day to the consumers. The RTP signal is a key feature of a smart meter which benefits the end user as
well as the utility. The RTP signal is provided to the users via smart energy meters, the users modify
their daily needs according to the RTP signal, and the EMC scheduled the smart appliances according
to the proposed algorithm. The day-ahead energy price signal of the New York Independent System
Operator (NYISO), shown in Figure 2, accessed on 27th December 2019, is reproduced and used for
cost calculation [32].
Electronics 2020, 9, 406 7 of 17

35
DAP

Electricity price (cents) 30

25

20

15
9 11 13 15 17 19 21 23 1 3 5 7
<----am---><------------------pm-----------------><---------am--------->
Time(hours)

Figure 2. Day-ahead pricing (DAP) reproduced signal [32].

5. Proposed Dragonfly Algorithm

To achieve low cost, low PAR, and minimum waiting time, we have proposed DA, brief details of
which are given below:
A dragonfly is a decorative insect, its scientific name is Anisoptera and belongs to kingdom
Animalia, and it is classified to phylum Arthropoda. The average life span of a dragonfly is six months.
The dragonfly insect is found in 3000 different species around the world [33]. There are two main
phases in the life-cycle of a dragonfly, which includes: nymph and adult. The dragonfly spends most
of his lifespan as a nymph and then become an adult after passing the metamorphism stage as shown
in Figure 3.

(a) (b)

Figure 3. (a) Real dragonfly, (b) Life-cycle of dragonfly [8].

Dragonflies are placed in the class of small predators; they rely on other small insects for their
survival. They also prey aquatic flies and even small fishes. The swarming nature is an interesting fact
of dragonflies. The purpose behind the swarming of dragonflies is: hunting and migration. Hunting is
static (stationary) swarm and migration is dynamic (traveling) swarm.
In the static behavior of swarm dragonflies in small groups over a specific area to make prey of
all other flying insects such as butterflies, mosquitoes, and many other small insects [34]. On other
hand, in a dynamic swarm, a large number of dragonflies migrate from one place to another place
over a long distance for finding a best habitat for their living [35].
The mentioned two swarming behaviors are similar to the main phases of optimization using
meta-heuristics: exploration and exploitation. The static swarm is the main goal of exploration, while
a dynamic swarm is favorable in the exploitation phase. The two swarming styles are briefly explained
and mathematically represented in the next section.
• Separation represents the static collision prevention of dragonflies in the swarm from other
dragonflies of the nearby vicinity.
Electronics 2020, 9, 406 8 of 17

• The alignment shows the velocity matching of one dragonfly in the swarm to the other individual
dragonfly in the same swarm of dragonflies.
• Cohesion represents the struggle of dragonflies toward the center of the mass of the nearby
individual dragonflies.

All of the individual dragonflies should attract themselves towards the food sources and prevent
their selves from the enemies to survive, which is the main goal of the swarming nature of dragonflies.
In consideration of these two behaviors, five main position updating factors are shown in the
equation below:
Separation can calculated by the following equation:

N
Si = − ∑ Z − Zk (8)
k =1

where Z represents the position of the current dragonfly, while Zk represents the position of a k-th
nearby dragonfly, and N is the number of all other nearby individual dragonflies.
Alignment can calculated by the following equation:

∑ Nk = 1Vk
Ai = (9)
N
where Vk is the velocity of a k-th dragonfly nearby.
Cohesion can calculated by the following equation:

∑ Nk = 1Xk
Ci = Z (10)
N
where Z shows the position of the current dragonfly, while Zk represents the position of the k-th nearby
dragonfly and N is the number of all other nearby individual dragonflies.
Attraction toward the food can be calculated by the following equation:

Fi = Z + − Z (11)

where Z shows the position of the current dragonfly , and Z + represents the position of the target food.
Distraction from the enemy can be calculated by the following equation:

Ei = Z − + Z (12)

where Z shows the position of current dragonfly and Z − represents the position of the enemy.
Two vectors are used for updating the position of dragonflies and simulations of their movement,
in which two vectors are: step (∆Z ) and position (Z). ∆Z represents the direction of the motion of
dragonfly and the step vector is represented mathematically by the following equation:

∆Zt+1 = (sSepi + aAligi + cCohi + f Foodi + eEnemyi ) + w∆Zt (13)

where (s, a, c, f , e, w) are the swarm factors during an optimization process.

The step vector is represented mathematically by the following equation:

Zt + 1 = Zt + δZt + 1 (14)

In the two equations above, t shows the present iteration.

For the arbitrariness, stochastic nature and exploration of dragonflies, the flying over the search
area is necessary, using an arbitrary walk (levy flight) in case there is no nearby solution for finding
Electronics 2020, 9, 406 9 of 17

the position of individual in the swarm; for solving this issue, the position of an individual can be
calculated by the following equation:

Zt+1 = Zt + Levy(d) × Zt (15)

In the above equation, t shows the present iteration, and d is the dimension of the position vector.
The levy flight can be found by the following equation:

c1|c2| β
Levy( Z ) = 0.01 × 1σ (16)
×
where c1 and c2 are two randomly selected constants between 0 and 1; β is also a constant number that
is selected according to the situation of the problem. The step by step process of the proposed DA is
depicted in Algorithm 1 i.e., Pseudocode and Figure 4 i.e., flow-chart:

Algorithm 1: Pseudocode of the proposed Dragonfly Algorithm

1 Parameters initialization: DA parameters, maximum size of the pop. , number of iterations.
2 Input: DAP As , Ans , as , ans , ρ, Xbat , λs , λns Random generation of the initial population of Dragonflies.
3 Fitness function is evaluated
4 Position of individual dragonfly is updated
5 while Numbers of iterations < population size do
6 for i =1: P do
7 for j =1: Q do
8 end
9 New solution is evaluated;
10 Present Pbest is assigned to the old Pbest
11 end
12 end
13 Output: ϕα Totals , ϕα Totalns , ϕα Total

Start
No
No. of iterations
>
population size

Initialize all the DA parameters

S,A,C,F,E etc

Yes

Initialize the random No

population of dragonflies Check whether best position of
dragonfly is achieved

Fitness function Yes

evaluation
Update old best position of food
No by new best position of food

Yes

Position of individual
End
dragonflies is updated

Figure 4. The Flowchart of Step by step process of the DA.

6. Results and Discussion

Simulations are conducted for achieving our objectives to check the effectiveness, productiveness,
and validity of our proposed algorithm. The central aim of our simulations is to show the optimal
Electronics 2020, 9, 406 10 of 17

scheduling for both possible scenarios, i.e., single smart home for single and thirty days and, secondly,
for thirty smart homes for single and thirty days. Due to the random nature of the heuristic techniques,
we have considered the results after average of 50 runs. In each home, we considered 12 smart
appliances with different lifestyles. They may have different LOTs and different appliances’ power
ratings. All appliances of our assumed cases are tabulated and briefly explained in Table 1 along with
their parametric values and types. For example, appliances may be of shiftable or non-shiftable class.
Only shiftable appliances are taking part in the scheduling process, while non-shiftable appliances
do not take part in scheduling process, as they must complete their operational time according to
their fixed allotted time slots. The operational time for refrigerator is from 8:00 a.m. to 7:59 a.m. on
the next day and that of interior lightening is 6:00 p.m. to 12:00 a.m. These two appliances must
complete their operational time in these hours without interruption and shifting. The simulation
results consist of daily basis hourly load, hourly cost, total cost, waiting time, and PAR for single and
thirty homes, respectively.

6.1. Daily Basis Hourly Load

Figure 5 depicts the daily basis hourly load curve for a randomly selected single home and
thirty homes for 24 h. According to the utility provided DAP signal, shown in Figure 2, peak hours
range from 4:00 p.m. to 8:00 p.m., during which the energy prices are high. It is therefore clear from
Figure 5a,b that both algorithms, i.e., GA and DA, have shifted the load from high price hrs to low
price hrs. This shifting has drastically reduced the consumed energy cost. In comparison to GA, DA
shows a bit of a uniform response.

Per hour load for single home

10
Unscheduled
GA Scheduled
8
Dragonfly Scheduled
Hourly load (kw)

0
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1 2 3 4 5 6 7 8
Time (h)

(a)
Hourly load for 30 homes
30
Unscheduled
25 GA Scheduled
Dragonfly Scheduled
Hourly load (kWh)

20

15

10

0
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1 2 3 4 5 6 7 8
Time (h)

(b)
Figure 5. Daily basis hourly load for un-scheduled and GA and DA scheduled (a) single home;
(b) 30 homes.
Electronics 2020, 9, 406 11 of 17

6.2. Daily Basis Hourly Cost

Figure 6 shows the hourly cost for a randomly selected single home and thirty homes for one day.
It is clear from both Figure 6a,b that, in the un-scheduled case, prices are high. However, due to the
shifting of the load from ON-peak hours to OFF-peak hrs, both GA and DA have reduced the cost for
single as well as thirty homes.

Per hour cost for single home

2
Un−scheduled
GA Scheduled
Horly electricity cost ($)

Dragonfly Scheduled
1.5

0.5

0
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1 2 3 4 5 6 7 8
Time (h)

(a)
Per hour cost for thirty homes
7
Unscheduled
6 GA Scheduled
Dragonfly Scheduled
Hourly electricity cost ($)

0
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 1 2 3 4 5 6 7 8
Time (h)

(b)

Figure 6. Hourly cost for un-scheduled and GA and DA scheduled (a) single home; (b) 30 homes.

6.3. Total Average Cost

Figure 7 shows the total average cost for a randomly selected single home for one day, single home
for thirty days, thirty homes for single day and thirty homes for thirty days. Due to the shifting of load
from ON-peak hours to OFF-peak hours, it is clear from Figure 7a that GA has 30.54%, while DA has
reduced the cost for a single home 35.57% for a one-day scenario. Similarly, in the case of a single home
for thirty days, Figure 7b GA and DA reduced the cost at 11.61% and 22.32%, respectively. Figure 7c
shows the case for thirty homes for a single day, while Figure 7d gives the results of thirty homes for
thirty days. Their respective values are given in Table 2.
Electronics 2020, 9, 406 12 of 17

Table 2. Results summary of Un-Scheduled, Genetic algorithm (GA), and Dragonfly algorithm (DA)
based scheduled load.

Techniques No. of Homes No. of Days Cost ($) % Cost Reduction Waiting Time (h) PAR % PAR Reduction
1 1 2.423 – – 4.62 –
Un- 1 30 36.017 – – — –
Schedule 30 1 48.142 – – 3.03 –
30 30 1432.735 – – — –
1 1 1.683 30.54% 3.03 3.56 22.94%
GA 1 30 31.832 11.61% — — —
Scheduled 30 1 44.982 06.56% 1.82 2.93 03.30%
30 30 1357.722 05.23% — — —
1 1 1.561 35.57% 2.89 3.76 18.61%
DA 1 30 27.977 22.32% — — —
Scheduled 30 1 39.851 17.22% 2.29 2.24 26.07%
30 30 1267.426 11.54% — — —

6.4. Daily-Basis 30 Days Load Pattern

Figure 8 depicts the daily-basis 30 days load response. It is clear from the figure that, in all three
cases (i.e., un-schedule, GA and DA scheduled cases), the total daily load remains same. It means
that, due to scheduling mechanisms, load does not reduce; it only shifts to OFF-peak hrs for reduction
of cost.

6.5. PAR
Figure 9 depicts the PAR of single home and 30 homes. It is clear from Figure 9a that GA has
reduced PAR to 22.94% for single homes and 03.30% for thirty homes, while DA has reduced it to 18.61%
and 26.07% for single and thirty homes, respectively, as shown in Figure 9b. Thus, GA performance in
terms of PAR reduction is better than DA; therefore, on the basis of simulation results, it is clear that,
in a single home scenario, GA will be preferred, while DA will be preferred for multiple homes scenario.

Total cost for Single home for one day Total cost for Single home for 30 days
2.5 40

2
30
Cost ($)
Cost ($)

1.5
20
1
10
0.5

0 0
Un
schedule GA Schedule Dragonfly Schedule Un−schedule GA Schedule Dragonfly Schedule

(a) (b)
Total cost for 30 homes for 1 day Total cost for 30 homes for 30 days
50 1500

40
1000
Cost ($)
Cost ($)

30

20
500
10

0 0
Unschedule GA Schedule Dragonfly Schedule Unschedule GA Schedule Dragonfly Schedule

(c) (d)
Figure 7. Total cost for un-scheduled, GA and DA scheduled load (a) total cost for a single home for
one day; (b) total cost for a single home for 30 days; (c) total cost for 30 homes for one day; (d) total
cost for 30 homes for 30 days.
Electronics 2020, 9, 406 13 of 17

Daily load curve for 30 homes for 30 days

260
Unscheduled
GA Scheduled
240
Dragonfly Scheduled
Daily load (kWh)
220

200

180

160
0 5 10 15 20 25 30
Days

Figure 8. Daily load curve for un-scheduled and GA and DA scheduled load for 30 days.
PAR for single home PAR for thirty homes
5 4

4
3

3
PAR

PAR
2
2

1
1

0 0
Un−scheduled GA Scheduled Dragonfly Scheduled Unscheduled GA Scheduled Dragonfly Scheduled

(a) (b)

Figure 9. PAR for un-scheduled and GA and DA scheduled load (a) single home; (b) 30 homes.

6.6. Average Waiting Time

Figure 10 depicts the average waiting time of single and thirty homes. Figure 10a shows that,
in terms of waiting time, DA gives better results for single home cases, while Figure 10b depicts that,
for the thirty homes scenario, GA has less waiting time. It concludes that DA reduces cost but at the
cost of consumer discomfort. Thus, those consumers who prefer to reduce their electricity bill instead
of taking into consideration more waiting time may use DA and vice versa.

Average waiting time for single home Average waiting Time for 30 homes
3.5 2.5

3
2
Waiting Time (h)

2.5
Waiting time (h)

2 1.5

1.5 1
1
0.5
0.5

0 0
Un−scheduled GA Scheduled Dragonfly Scheduled Un−scheduled GA Scheduled Dragonfly Scheduled

(a) (b)

Figure 10. Average daily waiting time for un-scheduled and GA and DA scheduled load (a) single
home; (b) 30 homes.

7. Comparison and Limitations

7.1. Comparison
Table 2 depicts the comparison of proposed algorithm DA with GA and un-scheduled load for
single and thirty homes. Three objectives—cost, PAR, and waiting time—are tried to minimize with
both algorithms. In terms of cost reduction, DA outperformed GA, while, for waiting time, GA is
Electronics 2020, 9, 406 14 of 17

good. However, in terms of PAR, GA is superior for single homes, but DA is more efficient for
multiple homes.

7.2. Limitations
Obviously, no such ideal algorithm is available with no limitations. The main limitation of our
proposed algorithm is minimization of an electricity bill at the cost of user discomfort. Secondly,
this algorithm is not performing well in all scenarios of single homes and multiple homes. Simulation
results show that the meta-heuristic optimization technique DA reduces the cost by 35.57% and 17.22%,
while GA reduces the electricity cost by 30.54% and 06.56% in single and multiple home scenarios,
respectively. Thus, DA is better than GA for cost reduction. However, simulation results of PAR show
that GA reduces PAR by 22.94% and DA reduces it by 18.61% for single homes, while, for multiple
homes scenario, GA reduces PAR by 03.30% while DA reduces it by 26.07%. Thus, in terms of PAR,
for single home scenarios of the smart grid, GA is more efficient than DA, while, for multiple homes
scenario, DA is more efficient than GA.

8. Conclusions
In this paper, a novel meta-heuristic optimization appliances scheduling technique in a single
home and multiple (thirty in our case) homes is proposed and analyzed. We proposed a new
nature-inspired optimization algorithm, DA for achieving two main objectives; reduction of consumer
electricity bill and PAR, keeping in view a specified waiting time threshold because of appliances
scheduling process. DAP is used for cost calculation. We considered two classes of appliances,
shiftable and non-shiftable. In shiftable appliances, we assumed only ten appliances for applying our
proposed algorithms to check their performance, while, in a non-shiftable class, we only took two
appliances. We compared our results with GA, a well known optimization algorithm in this field to
check three fitness functions mentioned, i.e., minimization of the energy cost, PAR, and waiting time.
Perhaps, a lot of countries in this bio-sphere can fulfill their energy demand. However, using such
optimization algorithms, the existing grid system can be improved and their journey towards smart
grids, to further facilitate their customers, by integrating renewable energy sources can be achieved.
In addition, increasing energy generation is not only practicable, but pollution is also increasing with
increased emission of carbon, while using different fuels for energy generation. Therefore, the benefits
of such optimization algorithms are not only to reduce energy cost, but also to minimize pollution.
The simulation results depict that our proposed technique performed well in the case of PAR and
cost reduction. However, the limitation of such algorithms is that it reduces the electricity bill at
the cost of user discomfort in the shape of increased waiting time. In the future, multi-objective
algorithms will be proposed to reduce the electricity bill and PAR, while maintaining a high level of
user comfort. In addition, the proposed multi-objective algorithms will be tested on all three sectors
of residential, commercial, and industrial areas, to achieve more benefits for both the utility and
consumers. Therefore, newly proposed nature-inspired algorithms will be applied for this purpose.

Author Contributions: M.U., I.U., and I.H. did the written work for the whole manuscript. I.U. and I.H. proposed
the system model and did the revision. A.B. and M.N. did editing and developed methodology, while M.S. and
D.S. technically supervised the whole work and refined it. All of the authors equally contributed, edited, revised,
and approved the manuscript. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by Hankuk University of Foreign Studies, Yongin, Korea and Endicott College
of International Studies, Woosong University, Daejeon, South Korean research fund.
Acknowledgments: The authors are extremely thankful to Hankuk University of Foreign Studies, Yongin, Korea
and Endicott College of International Studies, Woosong University, Daejeon, South Korea for supporting this work.
Conflicts of Interest: The authors declare no conflict of interest.
Electronics 2020, 9, 406 15 of 17

Abbreviations
The following abbreviations are used in this manuscript:

ABC Artificial bee colony

ANN Artificial neural network
BFA Bacterial foraging algorithm
BPSO Binary particle swarm optimization
CSA Cuckoo search algorithm
DA Dragonfly algorithm
DG Distributed generation
DR Demand response
DSM Demand side management
EDE Enhanced differential evolution
ED Economic dispatch
EDTLA Enhanced differential teaching learning algorithm
EWA Earthworm algorithm
EMC Energy management controller
GA Genetic algorithm
GWO Grey wolf optimization
HSA Harmony search algorithm
LOT Length of operational time
MIP Mixed integer programming
MVPA Most valuable player algorithm
OTI Operational time interval
PAR Peak-to-average power ratio
PEV Plug-in electric vehicle
PSO Particle swarm optimization
PV Photo-voltaic
RES Renewable energy sources
RTP Real-time pricing
SG Smart grid
SH Smart home
SM Smart meter
TG Traditional grid
TLGO Teacher learning genetic optimization
WDO Wind-driven optimization
TLBO Teaching learning-based optimization

References
1. Hussain, I.; Aamir; Ullah, I.; Khan, N.; Riaz, M. Reliable and Secure Advanced Metering Infrastructure for
Smart Grid Network. In Proceedings of the 2018 International Conference on Computing, Electronic and
Electrical Engineering (ICE Cube), Quetta, Pakistan, 12–13 November 2018; pp. 1–6. [CrossRef]
2. Strbac, G. Demand side management: Benefits and challenges. Energy Policy 2008, 36, 4419–4426. [CrossRef]
3. Yang, J.; Zhang, G.; Ma, K. Matching supply with demand: A power control and real time pricing approach.
Int. J. Electr. Power Energy Syst. 2014, 61, 111–117. [CrossRef]
4. Hu, W.; Chen, Z.; Bak-Jensen, B. Optimal operation strategy of battery energy storage system to real-time
electricity price in Denmark. In Proceedings of the IEEE PES General Meeting, Providence, RI, USA,
25–29 July 2010; pp. 1–7.
5. Tascikaraoglu, A.; Boynuegri, A.R.; Uzunoglu, M. A demand side management strategy based on forecasting
of residential renewable sources: A smart home system in Turkey. Energy Build. 2014, 80, 309–320. [CrossRef]
6. Logenthiran, T.; Srinivasan, D.; Shun, T.Z. Demand side management in smart grid using heuristic
optimization. IEEE Trans. Smart Grid 2012, 3, 1244–1252. [CrossRef]
7. Khan, I.; Singh, D. Energy-balance node-selection algorithm for heterogeneous wireless sensor networks.
ETRI J. 2018, 40, 604–612. [CrossRef]
Electronics 2020, 9, 406 16 of 17

8. Mirjalili, S. Dragonfly algorithm: A new meta-heuristic optimization technique for solving single-objective,
discrete, and multi-objective problems. Neural Comput. Appl. 2016, 27, 1053–1073. [CrossRef]
9. Lu, R.; Hong, S.H.; Yu, M. Demand Response for Home Energy Management Using Reinforcement Learning
and Artificial Neural Network. IEEE Trans. Smart Grid 2019, 10, 6629–6639. [CrossRef]
10. Javaid, N.; Hussain, S.M.; Ullah, I.; Noor, M.A.; Abdul, W.; Almogren, A.; Alamri, A. Demand Side
Management in Nearly Zero Energy Buildings Using Heuristic Optimizations. Energies 2017, 10, 1131.
[CrossRef]
11. Khan, Z.A.; Zafar, A.; Javaid, S. Hybrid meta-heuristic optimization based home energy management system
in smart grid. J. Ambient. Intell. Hum. Comput. 2019, 10, 4837–4853. [CrossRef]
12. Wu, X.; Hu, X.; Yin, X.; Moura, S.J. Stochastic Optimal Energy Management of Smart Home With PEV Energy
Storage. IEEE Trans. Smart Grid 2018, 9, 2065–2075. [CrossRef]
13. Ullah, I.; Hussain, S. Time-Constrained Nature-Inspired Optimization Algorithms for an Efficient Energy
Management System in Smart Homes and Buildings. Appl. Sci. 2019, 9, 792. [CrossRef]
14. Ou, T.; Hong, C. Dynamic operation and control of microgrid hybrid power systems. Energy 2014, 66,
314–323. [CrossRef]
15. Hussain, H.M.; Javaid, N.; Iqbal, S.; Hasan, Q.U.; Aurangzeb, K.; Alhussein, M. An Efficient Demand Side
Management System with a New Optimized Home Energy Management Controller in Smart Grid. Energies
2018, 11, 190. [CrossRef]
16. 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.
[CrossRef]
17. Ullah, I.; Hussain, I.; Singh, M. Exploiting Grasshopper and Cuckoo Search Bio-Inspired Optimization
Algorithms for Industrial Energy Management System: Smart Industries. Electronics 2020, 9, 105. [CrossRef]
18. Singh, D. Mobility and Energy Efficient Mechanism for Low power devices to Support Global Healthcare
System. J. Theor. Appl. Inf. Technol. 2014, 66, 315–329.
19. Ou, T.-C.; Su, W.-F.; Liu, X.-Z.; Huang, S.-J.; Tai, T.-Y. A Modified Bird-Mating Optimization with
Hill-Climbing for Connection Decisions of Transformers. Energies 2016, 9, 671. [CrossRef]
20. Jamil, A.; Alghamdi, T.A.; Khan, Z.A.; Javaid, S.; Haseeb, A.; Wadud, Z.; Javaid, N. An Innovative Home
Energy Management Model with Coordination among Appliances using Game Theory. Sustainability 2019,
11, 6287. [CrossRef]
21. Singh, D.; Lee, H.J.; Chung, W.Y. An energy consumption technique for global healthcare monitoring
applications. In Proceedings of the 2nd International Conference on Interaction Sciences: Information
Technology, Culture and Human, Seoul, Korea, 24–26 November 2009; pp. 539–542.
22. Ullah, I.; Khitab, Z.; Khan, M.N.; Hussain, S. An Efficient Energy Management in Office Using Bio-Inspired
Energy Optimization Algorithms. Processes 2019, 7, 142. [CrossRef]
23. Asgher, U.; Babar Rasheed, M.; Al-Sumaiti, A.S.; Ur-Rahman, A.; Ali, I.; Alzaidi, A.; Alamri, A. Smart Energy
Optimization Using Heuristic Algorithm in Smart Grid with Integration of Solar Energy Sources. Energies
2018, 11, 3494. [CrossRef]
24. Aslam, S.; Bukhsh, R.; Khalid, A.; Javaid, N.; Ullah, I.; Fatima, I.; Hasan, Q.U. An efficient home energy
management scheme using cuckoo search. In Proceedings of the International Conference on P2P, Parallel,
Grid, Cloud and Internet Computing, Barcelona, Spain, 8–10 November 2017; pp. 167–178.
25. Tsui, K.M.; Chan, S. Demand response optimization for smart home scheduling under real-time pricing.
IEEE Trans. Smart Grid 2012, 3, 1812–1821. [CrossRef]
26. Ramli, M.A.; Bouchekara, H.; Alghamdi, A.S. Efficient Energy Management in a Microgrid with Intermittent
Renewable Energy and Storage Sources. Sustainability 2019, 11, 3839. [CrossRef]
27. Zhao, J.; Liu, S.; Zhou, M.; Guo, X.; Qi, L. Modified cuckoo search algorithm to solve economic power
dispatch optimization problems. IEEE/CAA J. Autom. Sin. 2018, 5, 794–806. [CrossRef]
28. Nimma, K.S.; Al-Falahi, M.D.A.; Nguyen, H.D.; Jayasinghe, S.D.G.; Mahmoud, T.S.; Negnevitsky, M. Grey
Wolf Optimization-Based Optimum Energy-Management and Battery-Sizing Method for Grid-Connected
Microgrids. Energies 2018, 11, 847. [CrossRef]
29. Iqbal, Z.; Javaid, N.; Iqbal, S.; Aslam, S.; Khan, Z.A.; Abdul, W.; Almogren, A.; Alamri, A. A Domestic
Microgrid with Optimized Home Energy Management System. Energies 2018, 11, 1002. [CrossRef]
Electronics 2020, 9, 406 17 of 17

30. Al-Sakkaf, S.; Kassas, M.; Khalid, M.; Abido, M.A. An Energy Management System for Residential
Autonomous DC Microgrid Using Optimized Fuzzy Logic Controller Considering Economic Dispatch.
Energies 2019, 12, 1457. [CrossRef]
31. 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.
[CrossRef]
32. Available online: http://www.energyonline.com/Data/GenericData.aspx?DataId=2&${ISO-NE}_{---}{Day-
Ahead}_{-}{Energy}_{-}Price$ (accessed on 27 December 2019).
33. Beaudrot, L.; Ahumada, J.A.; O’Brien, T.; Alvarez-Loayza, P.; Boekee, K.; Campos-Arceiz, A.; Eichberg, D.
Standardized assessment of biodiversity trends in tropical forest protected areas: The end is not in sight.
PLoS Biol. 2016, 14, e1002357. [CrossRef]
34. Wikelski, M.; Moskowitz, D.; Adelman, J.S.; Cochran, J.; Wilcove, D.S.; May, M.L. Simple rules guide
dragonfly migration. Biol. Lett. 2006, 2, 325–329. [CrossRef]
35. Russell, R.W.; May, M.L.; Soltesz, K.L.; Fitzpatrick, J.W. Massive swarm migrations of dragonflies (Odonata)
in eastern North America. Am. Midl. Nat. 1998, 140, 325–343. [CrossRef]

c 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).