This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Date of publication xxxx 00, 0000, date of current version xxxx 00, 0000.
Digital Object Identifier 10.1109/ACCESS.2017.DOI

An Optimal Power Usage Scheduling in

Smart Grid integrated with Renewable
Energy Sources for Energy Management
ATEEQ UR REHMAN1 , ZAHID WADUD 1 , RAJVIKRAM MADURAI ELAVARASAN 4 , GHULAM
HAFEEZ 2,3,* , IMRAN KHAN 2 , (Senior Member, IEEE), ZEESHAN SHAFIQ 2 , and HASSAN
HAES ALHELOU4 , (Member, IEEE)
1
Department of Computer System Engineering, University of Engineering and Technology, Peshawar 25000, Pakistan;
2
Department of Electrical Engineering, University of Engineering and Technology, Mardan 23200, Pakistan;
3
Department of Electrical and Computer Engineering, COMSATS University Islamabad, Islamabad 44000, Pakistan;
4
Clean and Resilient Energy Systems (CARES) Laboratory, Texas A&M University, Galveston, TX 77553,USA;
5
Department of Electrical Power Engineering, Faculty of Mechanical and Electrical Engineering, Tishreen University, Latakia 2230, Syria;
Corresponding authors: Hassan Haes Alhelou; Ghulam Hafeez (e-mail: [email protected]; [email protected])
This work was supported by Tishreen University, Latakia 2230, Syria.

ABSTRACT Existing power grids (PGs) and in-home energy management controllers do not offer its
users the choice to maintain comfort and provide a bearable solution in terms of low cost and reduced
carbon emission. This work is based on energy usage scheduling and management under electric utility
and renewable energy sources i.e., solar energy (SE), controllable heat and power (CHP) and wind energy
(WE) together. Efficient integration of renewable energy sources (RES) and battery storage system (BSS)
have been suggested to solve the energy management problem, reduce the bill cost, peak-to-average ratio
(PAR) and carbon emission. User’s electricity bill reduction have been achieved by proposed power usage
scheduling method and integrating low cost RESs. PAR minimization have been achieved through shifting
the demand in response to real time price from high-peak hours to low-peak hours. In this context, load
scheduling and energy storage system management controller (LSEMC) is proposed which is based on
heuristic algorithms i.e., genetic algorithm (GA), wind driven optimization (WDO), binary particle swarm
optimization (BPSO), bacterial foraging optimization (BFO) and our suggested hybrid of GA, WDO and
PSO (HGPDO) algorithm. The performance of the heuristic algorithms and proposed scheme is evaluated
numerically. Results demonstrate that our proposed algorithm and the LSEMC reduces the electricity bill,
PAR and CO2 in Case 1, by 58.69%, 52.78% and 72.40%, in Case 2, by 47.55%, 45.02% and 92.90% and
in Case 3, by 33.6%, 54.35% and 91.64%, respectively as compared with unscheduled. Moreover, the user
comfort by our proposed HGPDO algorithm in terms of delay, thermal, air quality and visual improves by
35.55%, 16.66%, 91.64% and 45%, respectively.

INDEX TERMS Energy management; battery energy storage systems; renewable; hybrid heuristic
algorithms; power usage scheduling; smart grid.

I. INTRODUCTION tration (EIA), the average electric bill for U.S. households
could increase by 2.3% next year [3]. This drastic increase
With the growth of population and development, it is esti- in demand and cost will result int the need to generate
mated that the energy demand may grow by 3% at the end electricity from alternative sources like solar, thermal, and
of year 2021 [1]. Traditional power grids (PGs) powered by wind. Moreover, to cover the exponential increase in energy
fuels, produce 64.5% of power worldwide. These PGs emit demand, reduce carbon emission and low cost, researchers
a larger amount of carbon, where the generation sector and have suggested new means of power generation by renewable
transport sector almost release 40% and/ 24% carbon respec- energy sources (RES) [4]. To efficiently use these RES, we
tively [2]. According to the Energy Information Adminis-

VOLUME 4, 2016 1

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An optimal power usage scheduling in smart grid

need to renovate traditional PGs into smart grids (SGs). SG follows:

has the ability to meet growing demands and incorporate • Efficient Integration of power storage systems and gen-
new RES together. SG incorporate modern communication eration systems i.e., BSS, solar, thermal, WE, electric
infrastructure with existing grid to efficiently use available vehicles storage system (EVSS) and external electric
energy sources at site [5]. grid (EEG) have been suggested to solve the problem
Nomenclature of energy management.
Main symbols
vjt Velocity of particles at t
• The SE, CHP, WE, BSS, EVs and domestic consump-
Fc
µ
Coriolis force
Inertia factor
tion is made controllable so that the management of
Ω
αc
Earth rotation
coefficient of friction
power is possible.
ρ
Ff
Air density
Friction force
• To reduce power consumption form EEG in response to
Fpg Friction gradient RTP which narrows and shifts load in peak hours.
Fg Gravitational force
δv Finite volume of the air • Procurement of energy from low price systems i.e. SE,
∆ Pressure gradient
z Ant CHP and wind to reduce the user electricity bill.
g Acceleration due to gravity
v Smart home • Efficient BSS integration and utilization of EVSS in
x Set
M Set of shift-able appliances peak hours.
T Time interval
t Time slot • Maximize user comfort (UC) by minimize delay (delay
N Set of non-shiftable appliances
vit Current velocity at t is the waiting duration to serve the appliances).
Psch Power scheduling matrix
A All appliances • PAR and CO2 emission percentage reduction.
DT R Delay time rate
Ngmax Maximum iteration • Average UC maximization in terms of delay, visual
ρDT Ra
X
t−1
Total DTR
Global best position
comfort (VC) and thermal comfort (TC).
gbest
Eex (t) Energy purchased from external grid • Scheduling HVAC which uses almost 50% of the load
Lt Total load
Lsch
bill Schedulable load bill [25], in such a way to not compromise the TC and
Lnsch
bill Non-schedulable load bill achieve less delay comfort and bill as compared with
Epv PV power
Pwd
Ebio
Wind generation power
CHP generation power
the unscheduled.
l Appliance operation time
α/β Operation start/end time The remaining work is organized as follows: section II
η Price per kWh
γ Electricity emissions factor presents the related works, section III explains the proposed
φ(t) Total RES at time t
Epv Energy conversion efficiency factor algorithms, section IV describes the proposed system model
Apv Area of panel in (m2 )
Ir(t) Solar irradiance at time slot t and, section V discusses the simulation results. Section VI
Ta (t) Outdoor temperature at time slot t
ζ Weighted factor concludes the work.
α1 , α 2 Shape factors
β1 ,β2 Scale factors
Atb Area of wind turbine blade
Vs (t) Wind speed in m/s at time slot t
∂ Total system efficiency
We
Qth
CHP electric output
Total thermal output
II. RELATED WORKS
L
Qf
Lower heating value
Fuel input to CHP
Several approaches have been adopted for optimized energy
Qbr
γe
Bio-gas recovery
Recovery efficiency
management and load scheduling. Moreover, with the advent
CF
pt
Capacity factor
EV on-board PV generation
of information technology, the demand for power rises day-
gt
lτ
Remaining energy of pt
EV load at time slot τ
by-day. In the literature, for enhanced energy management
ω
ηe
Time constant of HVAC system
Thermal conversion efficiency
and load scheduling, numerous techniques have been sug-
Ac
Abbreviations
Overall thermal conductivity gested. Providing opportunity for the user to schedule and
UC, TC, VC User , thermal and visual comfort shift their demand from high peak hours to off peak hours
NSA Non-schedulable appliances
LOT Lyapunov optimization technique using energy management controllers (EMC). Related work
SSA Smart schedulable appliances
BPSO, PSO Binary/particle swarm optimization with techniques used, objectives and limitations are summa-
PAR Peak to average ratio
EP Electricity price rized in Table 1.
RTP Real time price
GA Genetic algorithm With the SG advancement, a user can reduce the electricity
WDO Wind driven optimization
HEMS Home energy management system bill charges by integrating the RESs. In [6], the authors
SG, MG Smart grid, micro grid
HGPDO Hybrid-GA, PSO and WDO discussed an artificial neural network (ANN) based model for
the integration of RESs to reduce cost and minimize carbon
At user end, there is absence of energy management emission. As a result the energy cost of the consumer is
and load scheduling. Moreover, at traditional PGs, there is reduced by 35%. However, they ignored the integration of
lack of communication infrastructure. Besides, due to excess BSS and user comfort in their work. In [7], the authors sug-
emission of carbon from fuel based PGs, a rise in the en- gested both GA and WDO and then compared the results. The
vironmental pollution has occurred [32]. It is not feasible results showed 29% and 36.2% reduction in the bill cost and
for the consumers to change the schedule of their electric PAR respectively. However, they have ignored RESs integra-
appliances by compromising comfort level. Accordingly, the tion in their work. The authors in [8] suggested the techniques
LSEMC is required, which smartly schedule the load of of Harris’ Hawk optimization combined with integer linear
home appliances. The main highlights of our research are as programming (ILP) to solve the randomness problem and to

2 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An Optimal power usage scheduling in smart grid

schedule the user appliances. Authors’ main objectives were presented a storage management model by installing EV
to analyze cost and the trade-off strategies for user comfort batteries as backup. It further involves integration of EV
and financial benefits. Their model is adoptable to user batteries system to smart grid by adopting MILP formula-
requirement and robust levels, however, carbon reduction tion. However, battery degradation and capacity were not
and RES has been ignored. In [9], to solve the optimization included in cost reduction. In [22], authors particularize a
problem, minimize operation cost and carbon emission, the demand side-management methodology that is useful for
authors suggested techniques of multi-objective genetic algo- households based on ANN. Their model includes PV and
rithm (MOGA). The authors in [10] have considered a smart energy storage and aim a decision-making system to reduce
home, which is connected to a grid while different appliances bill, however, battery degradation cost and user comfort is
were drawing energy from external grid. Integration of mix not considered. In [23], authors suggested a home energy
energy system has been discussed which includes energy management system model to establish and predict a con-
storages, renewable sources and photo voltaic (PV). Authors trol system based on mixed-integer quadratic programming
in [11] have used a multi-objective linearization approach (MIQP) technique powered by thermal source. Their main
for home energy management with RES and plug-in hybrid objective was to curtail electricity bill, though RESs were
electric vehicle (PHEV). Their objectives were to minimize not integrated. In [24], authors have suggested micro-grid
energy bill and load profile deviation. Nevertheless, user concept to run data centers on it, to mitigate carbon emission,
comfort has not considered. In [12], authors have proposed reduce bill and power outage issues. They used lyapunov
an intelligent approach for demand side management with optimization technique (LOT) to design and simulate their
forecasting and net-metering structure. Their objectives were system model. In [25], an energy management system for
to reduce cost, PAR and carbon emission. In [13], authors sustainable smart home is suggested with HVAC load and
have developed a load scheduling and energy management random occupancy. Their objectives were to minimize cost
model for electric vehicle, which was connected to external and reduce cost of thermal discomfort using LOT. Although,
grid and charging stations. In remote areas, where power carbon emission reduction in system model were ignored.
grid extendibility is costly and not feasible, the mini grid In [26], a novel controller for scheduling of appliances and
concept is effective and reliable [14]. The authors in [13]- integrating RES in virtual plant have been designed. They
[15] have worked on management of storage and real-time have used PSO algorithm to minimize cost in crucial time and
load scheduling of renewable energy (RE) integrated electric prioritize the sustainable resources. In [27], to increase the
vehicle (EV). Their main objectives were to minimize energy smart grid potential, they have integrated smart homes with
gaining from external grid and charge stations which is park- renewable energy. The main objectives were PAR and cost
ing station, public and home charge stations. Moreover, delay reduction, but the impact of comfort were not considered. In
minimization, load scheduling, price reduction of aboard [28], a data focused machine learning (ML) approach applied
PV, PV efficiency and reduction of battery degradation cost to model the demand and electricity forecasting have been
and yearly carbon in air were also their objectives. In [16], proposed. However, EMC was not included in this work. In
authors have proposed a model for IoT-enabled smart home [29], a multi-headed CNN based model for price prediction
for energy management and load scheduling. Their main and integration of RES and BSS with SG were suggested.
objectives were to alleviate peak formation and reduce the Their proposed scheme helps to reduce the electricity bill of
cost with constraints of user comfort. In [17], an innovative users by 58.32% and 63.02% through integration of RESs
home EMC based on ANN and day-ahead grey wolf modified without and with BSS, respectively. In [30], authors have
enhanced differential evolution algorithm (DA-GmEDE) is proposed a dynamic programming (DP) based EMC for
suggested. They develop a strategy for energy management shifting load of demand side management to optimize the
with day-ahead demand response and power consumption smart home appliances usage pattern. Their objectives were
forecasting in SG. In [18], authors have recommended a to reduce cost, PAR and maintain UC level. In [31], authors
smart charge and discharge scheduling algorithm based on have presented a domestic demand model, which showed all
linear programming (LP) for EV. However, user comfort, dwelling appliances scheduling pattern for twenty two homes
carbon emission and RE integration to grid were ignored. over a year. In [32], a hybrid programmable home man-
In [19], authors have suggested a nano-grid concept to cater agement system has been proposed based on PSO and GA
peak demand and reduce blackouts. This model work for algorithm. However, thermal comfort and CHP generation
residential side to auto disconnect the low priority loads. was not included. In [34], authors have suggested an optimal
Their objectives were to reduce system from blackout based home EMC based on heuristic algorithms. Their objectives
on the neighbour level. However, user comfort is not included were to reduce cost, PAR and integrate RES into the system
in system model. In [20], authors have presented a pyramid model. They have achieved 59.06% and 17.40% reduction
convolution neural network (CNN) based learning model for in bill cost and PAR respectively. In [35], an optimal load
energy forecasting. They have suggested that power con- scheduling framework based on hybrid gray wolf-modified
sumers can be grouped into clusters and then representative enhanced differential evolutionary (HGWmEDE) algorithm
system could be designed and trained, which can accurately has been suggested. They scheduled the household load using
forecast power load for each customer. In [21], authors have the output of the forecaster module. In [36], a shaping load

VOLUME 4, 2016 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An optimal power usage scheduling in smart grid

techniques based on demand side management for industrial In the above equation u is the inertia factor, vit and vjt is
plant has been optimized using GA. It has achieved an overall current velocity and velocity of particle, respectively. k1 and
PAR reduction of 21.91%. k2 are random numbers while, z1 , z2 are local and global pull,
respectively. x is the particle’s current position, Xlbest and
III. PROPOSED MODEL Xgbest are the local and global best positions respectively. To
In this section, our proposed model is discussed in detail. map the velocities of particle between 0 and 1, equation (2)
The methods which have been adopted have also been sum- is used.
marized. We used BPSO, GA, WDO, BFO and HGPDO, 1
sim vit+1 (j) =


because the said algorithms have heuristic nature and initial- (2)
1 + exp(−vit+1 (j))
ization of population, which initially tends to optimal and
best solutions and subsequently, fills up to the rest of the C. WDO
population with the random solutions. Previously, LP, ILP, WDO is a heuristic based optimization algorithm. It is
DP, MILP, etc., methods have been adopted for appliances centered on air particles motion phenomenon in atmo-
scheduling and multi-objective problems. Nevertheless, these sphere. In this technique, an N-dimensional exploration space
methods cannot handle a bulky number of home appliances is formed, in which unlimited particles of air move. In
and also face several convergence difficulties. GA, BPSO, WDO, frictional, gravitational, Coriolis, and pressure gradi-
WDO, HGPDO, BFO algorithm overtakes the classical tech- ent forces are involved to control air particles. Each force
niques of optimization, and for solving multi-objective prob- have their own functions i.e., pressure gradient force and fric-
lems it provide various methods. The parameters and their tion force function for shifting the air particles in the forward
respective values are specified in Table 2. direction and resisting this forward direction respectively.
Also, the force of gravitational function pulls the particles
A. GA of air in the direction of origin, and the Coriolis force’s act to
The GA is adopted which is a natural inspired algo- detect the particles of air in the atmosphere. To calculate the
rithm to find optimal solutions. It randomly generates so- frictional, gravitational, pressure gradient force and Coriolis
lutions of population containing a defined number of char- force, equations (3), (4), (5) and (6) are used, respectively.
acters/individuals. Each solution contains a set of all kinds Which are given below [32].
of variables denoted as a chromosome. New solutions can
be obtained which contains old and new characteristics by Fpg = −∆ρσv (3)
calculating the fitness values, after that selecting individuals, Fc = −2Ω × µ (4)
crossover and mutation. A solution of best fitness can be
generated after its judgement. The defined GA algorithm is Fg = −ρσv × g (5)
used as taken in [32]. Referring to the selection process, the
Ff = −ρσv (6)
roulette selection method is used, in which the individual
with a better fitness value has a higher probability to be Where Coriolis force; velocity factor of wind is represented
selected for further processing. In general, the chromosomes by Fc . The Ω notation represents the earth rotation. The
are exemplified through binary strings and these are taken symbols Ff and α denotes friction force and friction coef-
easy to splitting and recombining. Table 2 shows all parame- ficient respectively. Fpg is the gradient force due to pressure,
ters with values used in algorithm. δv represents the air finite volume, ∆ denotes the gradient
of pressure, Fg denotes the gravitational force, ρ and g is
B. BPSO the density of air and acceleration because of gravitational
BPSO is a technique inspired by nature for searching optimal force respectively. In WDO, particles of air are taken as n,
solution inside a search space. Initially, it was presented in and random solutions are formed from these particles. A
form of continuous domain. But, later it was explored to dis- fresh population is produced, after updating velocities and
crete domain. The PSO discrete domain is known as BPSO. evaluating the fitness function. After this, the comparison of
It has mainly dependency on four aspects: (i) individual best old and new air particles fitness function for an optimal home
position of particles, (ii) best global position, (iii) initial appliance scheduling structure is found.
velocity and (iv), the initial position amongst all the particles.
In BPSO, an exploration space is formed, and a randomly D. BFO
population is initialized and spread in the exploration space. BFO is a nature-stimulated optimization algorithm. It is
To update the particles velocities in each iteration equation inspired by the social searching behavior of Escherichia coli
(1) is used in [32]. While the parameters values are in Table bacteria. In BFO algorithm, the bacteria swim in exploration
2. of n number of nutrients and choose the best nutrients (solu-
24 X
X I X
24 X
I 24 X
X I tions) to make the most of its energy. BFO also has four steps:
vit+1 = uv ti (t) + reproduction, elimination-dispersal, swimming and chemo-
t=1 i=1 t=1 i=1 t=1 i=1
(1) taxis. In the chemotaxis step, the parameters initialization of
z1 k1 (Xlbest , i (j) + z2 k2 (Xgbest , i (j) − xti (j)) foraging starts, after that the parameters initialization takes

4 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An Optimal power usage scheduling in smart grid

Table 1: Related work summary.

Reference No Techniques Objectives Limitations/Research gaps

[6] ANN Integrate RESs to reduce bill and carbon emission BSS and UC is ignored
[7] GA,WDO Reduce bill and alleviate PAR RE integration is not included
[8] ILP Schedule appliances and reduce uncertainty No RES integration and carbon emission reduction included
[9] MOGA Reduce operation cost and CO2 emission No UC
[10] LOT Reduce cost and PAR No UC and carbon emission
[11] Multi-objective linearization technique Reduce cost, and load profile deviation No UC
[12] HBFPSO Curtail cost, PAR and carbon emission No WE,SE and CHP
[13]- [15] LOT Minimize system cost and carbon emission Applicable to remote area
[16] Hybrid-WDO, BFA Minimize cost, PAR and improve comfort No BSS model included and carbon emission is ignored
[17] DA-GmEDE Cost reduction, PAR reduction and UC enhancement Carbon emission is ignored
[18] LP EV smart charge/discharge schedule controller Cost and carbon emission is not discussed
[19] UFUV Energy management and load reduction No UC and carbon emission reduction included
[20] CNN Energy forecasting UC and BSS model ignored
[21] ANN PV integration and reduction bill Battery cost ignored
[22] Game theory PAR and cost reduction Low comfort
[23] ML Demand forecasting No EMC model
[24] NN Shape demand and estimate load UC is ignored
[25] LOT Improve comfort, reduce electricity bill UC is ignored
[26] PSO Integrating RES in virtual plant No BSS and UC included
[27] Analytical model To increase SG potential by integrating RES No UC included
[28] ML Demand forecasting No EMC model
[29] CNN Price prediction, RES and BSS integration UC is ignored
[30] MKP, DP Reduce electricity bill and PAR BSS model is ignored
[31] Stochastic, LOT Optimize power cost Thermal and visual comfort ignored
[32] HGPO Reduce bill, PAR and carbon emission Low load and no thermal comfort included
[33] Analytical model Large BSS integration to PV system High cost
[34] Hybrid of GA and ACO Reduce cost, PAR and carbon emission No wind generation and less UC achieved
[35] HGWmEDE Optimal load scheduling and handle load randomness No RES and BSS model included
[36] GA Shaping load and demand Comfort is not included

Table 2: Algorithms parameters

place, the primary positions of all appliances are examined,
and then the new positions of the bacteria (solution matrix) Algorithm Parameters Values
GA Number of iterations 200
are calculated by the scheme. In the second step, the swim- Population size 200
ming loop is initialized to find the current best condition of Pm 0.1
the appliances, after this step is completed, the reproduc- Pc 0.9
N 11
tion iteration loop starts. For the new population, only the BPSO Number of iterations 200
fittest solution is used. Finally, the minimum fit solutions are Swarm size 200
removed, and new random samples are fitted with reduced Vmax 4
Vmin -4
probability. This is an important step because, the least fit Wi 2
solutions are rejected, and the probability of recurrence is C1 0.4
minimized. C2 2
N 11
WDO Number of iterations 200
E. HGPDO Population size 200
The HGPDO is our suggested algorithm. In HGPDO, the dimmin -5
dimmax 5
features of GA, BPSO and WDO algorithms are combined to Vmin -0.3
proficiently reduce electricity bill cost, PAR and UC in terms Vmax 0.3
of VC, TC and CO2 emission. The GA, WDO and BPSO Rt 3
n 11
are chosen because these algorithms efficiently reduces PAR, g 0.2
effectively reduce electricity cost and return maximum UC in a 0.4
terms of visual and thermal comfort, respectively. HGPDO BFO Number of iterations 200
Ne 24
primarily has three stages where, in the stage one, the initial Nr 5
steps of PSO are implemented. In the stage two, the steps of Nc 5
WDO are adopted and in the third stage, the GA mutation Np 30
Ns 2
and crossover are applied to the current global best position Ci 0.01
t−1 T C,V C
Xgbest , form by BPSO and U Cbest found by WDO, Ped 0.1
respectively. The GA features of crossover and mutation θ 0.1
HGPDO Same parameters as Same values as of
functioned at the best global position and best comfort offers of GA, PSO and WDO GA, PSO and WDO
good results as compared with their working to the individual
and random based population. Figure 2 presents the flowchart
of HGPDO algorithm. The parameters used in HGPDO and
their values on which its best optimal results are achieved IV. SYSTEM MODEL
are listed in Table 2. All parameters and constraints values Consider a residential area, an EG connected smart home
of GA, PSO and WDO are applied. The proposed method building (SHB) is fitted out with smart meters having load
inputs, parameters initialization and step by step computation scheduling and energy storage system management con-
are given in Algorithm 1. troller (LSEMC) which employ and work on our proposed
HGPDO algorithm. The LSEMC fetch all elements, such

VOLUME 4, 2016 5

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An optimal power usage scheduling in smart grid

Figure 1: Proposed HGPDO algorithm flowchart

as electrical home appliances, user comfort signal (share respective constraints, (iv) micro-combine heat and power
preferences of appliances), renewable energy sources (RESs) (CHP) and gas boiler (GB) generation values and relevant
local generations i.e., SE, WE, CHP running on biomass constraints, (v) level of energy of both BSS and EVSS and
and energy storage devices i.e., home BSS and electric their relevant constraints, (vi) real time price signals of elec-
vehicle storage system (EVSS). Our algorithm will check tricity and external electricity grid constraints. On sharing the
for EV availability at home and having sufficient energy above information, the LSEMC apply the proposed algorithm
level to serve as backup at peak hours. Smart appliances to determine optimal scheduling, manage the energy and
will communicate with controller through home area network storage system. It is assumed that all the LSEMC operations
and LSEMC will control scheduling of those appliances on are performed at fixed time slots, carried per time slot, and
proposed scheme. all the essential data is securely and timely brought by a
A LSEMC is operated at the external grid connected SHB communications network. There are many wireless solutions
which gathers these information: (i) UC signal about the for communication purpose such as Wi-Fi, ZigBee, Z-Wave,
loads pattern (i.e., scheduled or urgent; in terms of their dura- or a wired home plug protocol in SG. A simple system
tions of operations, tolerable delay, time of arrival, etc.) and architecture of SHB is presented in Figure 2.
comfort in terms of optical, heat, ventilation and air controller
(HVAC), (ii) PV power generation output and corresponding
constraints, (iii) Wind power generation output values and

6 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An Optimal power usage scheduling in smart grid

Figure 2: Proposed system model

24 n
!
A. ENERGY PROCUREMENT X X
Lnsch Lnsch nsch


bill = n∈N (t) × Sn∈N (t) × EP (t)
The smart building in residential area procures energy for t=1 N =1
scheduled loads and urgent loads from EG, in home PV, WE (9)
24 m
and CHP. While in peak hours, it also uses BSS along with X X
the EV battery as backup if available at home. Ltotal sch nsch sch
bill = Lbill + Lbill = Lbill = Lsch
m∈M (t) ×
t=1 M =1
n
X
sch
B. EXTERNAL GRID Lnsch nsch


Sm∈M (t) × EP (t) + n∈N (t) × Sn∈N (t) × EP (t)
The smart building is purchasing energy from external grid N =1
at defined price, discussed in subsequent subsection. The  (10)
sch 1 if schedulable appliance is ON
building procure energy from the external electricity grid. Let Sm∈M (t) = (11)
0 if schedulable appliance is OF F
Eex (t), be the procured amount of energy from the grid at 
time t. nsch 1 if non − schedulable appliance is ON
Sm∈N (t) = (12)
0 ≤ Eex (t) ≤ Eex,max (7) 0 if non − schedulable appliance is OF F
sch
Where Sm∈M (t) represent the indication of on/off state of
C. PRICE MODEL nsch
M number SSA and Sn∈N (t) represent the on/off state of N
There are several electricity tariffs, to set a price over 24 number NSA, and EP (t) is electricity price in the particular
hours i.e., peak pricing (PP), time of use pricing (ToUP), time slot t. The electricity bill L at any time slot t after taking
critical peak pricing (CPP) and real time pricing (RTP) [32]. all RES and BSS into consideration is calculated as following
In most of the appliances scheduling systems, the pricing of
electrical energy is supposed to be ToUP or RTP. Moreover, Lbill = (Lsch (t) + Lnsch (t) − (Eex (t) + Epv (t) + Pwd (t) +
in ToUP, the pricing of total duration is divided into different Ebio (t)) − BSS(τ ))
slabs and for each slab a fixed price is defined. In this work, (13)
we practice RTP, in which the price of electrical energy vary Where the τ is the duration when the RESs are either not
on the hourly basis and rests constant during an hour’s dura- available or sufficient, so the load drains energy from BSS.
tion. Equations (8) and (9) calculate the 24 hours electricity Our main objective function is given in (14) below.
bill of smart schedulable appliances (SSA) Lsch and non- T
schedulable appliances (NSA) Lnsch respectively.
X

min Lbill (t) + CO2 + P AR + Delay (14)
t=1
24 m
!
X X
sch sch sch


Lbill = Lm∈M (t) × Sm∈M (t) × EP (t) subject to constraints (15)-(19).
t=1 M =1
(8) Lsch (t) + Lnsch (t) = (E (t) + BSS (t) + ϕ (t)) (15)

VOLUME 4, 2016 7

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An optimal power usage scheduling in smart grid

Algorithm 1: HGPDO Algorithm steps and β shows the appliance operation start and end time
A. Algorithm Inputs: RTP, operation time duration, energy usage pattern, wind speed in m/s, temperature, respectively. While, ϕ(t) denotes the total procurement of
solar irradiance, efficiency, bio-gas availability, prior-scheduling pattern of appliances, BSS, CHP and EVSS
initialization. renewable energies at time slot t.
B. Parameters initialization: Number of iterations, population size, Pm , Pc , N , Wi , C1 , C2 ,
dimmin , dimmax , Vmin , Vmax , Rt , n, g, a, maximum and minimum velocity, number
of swarms, global and local pulls, initial and final momentum weights, cost per hour, PAR, Tc , Vc and
emission of CO2 .
for Time=1:24 do
D. PV GENERATION SYSTEM
For initial position
for hour=1:swarm do PV panel solar power generation depend primarily on solar
for i=1:n do
if rand>0.7 then radiations and over-all estimated radiation [32]. Solar power
X=1
end output depend on radiation amount, direction of panels and
else
X=0 transfer efficiency [34]. The generated energy in each time
end
end slot within 24 hours is firstly supplied to scheduled load and
end
For initial velocity is characterized as in (20) [32].
for vel1=1:10 do
for vel2=1:6 do
if Thour=0 then
H=0
Epv (t) = Epv Apv Ir (t) (1 − 0.05) (Ta (t) − 25) ∀ t
end
end (20)
end
return Vbest1 , return Pbest1
In equation (20), Epv denotes the energy efficiency conver-
For initial velocity and initial position, Calculate V elbest1 and P osbest1
Generate population for WDO
sion factor of the solar panels, Apv is the surface area of the
Assign position and velocity
begin
panels (m2 ), Ir(t) is the radiance of solar (kW/m2 ) at time
Based on 1st position and velocity do evaluation of fitness function t, the correction factor of temperature is 0.005 [34], Ta (t)
Using equations (1) and (2) update the velocity and position respectively
V elbest2 , P osbest2
Based on 2nd best position and velocity do evaluation of fitness function
denotes the outside temperature in (°C) at time t and stan-
Using equations (3)-(6) update velocity, position, check limits and boundaries, respectively
Evaluate the fitness for WDO
dard room temperature is 25(°C). The sun hourly irradiation
Check for Tc and Vc feasibility:
if thermal comfort and visual comfort is less than WDO scheduled comfort then
distribution generally follows a bimodal spreading that can
Crossover wbest and gbest , Crossoverresult = crg be measured as a linear blend of two unimodal distribution
end
else functions. It could be demonstrated using Weibull probability
Crossover P osbest1 and P osbest2 , Crossoverresult = crg ,
Mutate crg , M utationresult =globalbest density function as in equation (21) [34].
end
end   (α1−1)
returned globalbest α1 Ir (t) Ir α1
C. Compute objectives: EBC, PAR, Carbon emission and UC f (Irpv (t)) = ζ e−( β1 ) + (1 − ζ)
begin
1. Compute user comfort
β1 β1
for returned globalbest , to calculate UC do   (α2−1)
a. Usage power schedule=globalbest α2 Ir (t) Ir pv α2
b. Compute TC using equations (29) and (43) e−( β1 ) 0 < Ir (t) < ∞
c. Compute VC using equations (39) and (41)
d. Compute delay comfort using equation (38)
β2 β2
end
e. Compute air quality comfort by equation (44) (21)
2. Compute electricity cost
for returned globalbest , to calculate cost do
Where, ζ denotes a weighted factor, α1 and α2 are the shape
a. Only EG: Lt = P ower×globalbest
b. EG with RES: Lt(RES)=Load-Energy from RES
factors, together with β1 and β2 which are scale factors.
c. EG with RES and BSS: Lt(RES and BSS)=Load-Energy procured from RES-BSS
discharge
d. Compute EBC by equation (13) for Case a, b and c
end E. WIND ENERGY
3. Compute CO2 emission
for returned global-best, to calculate CO2 emission do Power generation of wind turbines owed to the kinematic
a. Only EG: average cost excluding RES and BSS
b. EG with RES: average cost including RES energy of wind speed, hence, its production of electrical
c. EG with RES and BSS: average cost including both RES and BSS

end
d. Compute CO2 by equation (37) for Case a, b and c energy primarily depends on meteorological conditions and
4. Compute PAR
for Operated returned global best, to calculate PAR do
direction of wind flow. It is assumed that procured amount of
a. Only EG: Lt = P ower×globalbest
b. EG and RES: Lt(RES)=Load- Procured energy by RES
energy from wind turbine is given by:
c. EG, RES and BSS: Lt(RES and BSS)=Load- Procured energy from RES-BSS
discharge 1 3
d. Compute PAR by equation (36) for Case a, b and c Pwd (t) = ∗ ρ ∗ Atb ∗ Vs (t) (22)
end
end 2
end
Where, ρ, Atb denotes the air density and area of turbine
blade, respectively, while, the symbol Vs (t) denotes the air
speed in m/s.
n
X
η = l(a) (16)
F. MICRO-CHP
a=1
In order to meet the SHB power demand, it is supposed
n
X that the building is acquiring energy from its own control-
α≤η≤β (17)
lable micro-CHP generation system. Details of CHP system
a=1
are specified as follows. CHP system electricity generation
ϕ (t) = RES (18) on average uses 32% less fuel, thus results in 50% less
carbon emission. Renewable fuel such as biomass, biogas,
renewable natural gas (RNG) and renewable hydrogen (RH)
0 ≤ BSS min ≤ BSS max , ∀t ∈ T (19)
have potential to reduce carbon emission even further. For
Where, l represent length of appliance operation time, α thermal generation, electrical generation or as transport fuel,

8 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An Optimal power usage scheduling in smart grid

the methane CH4 biogas can be used [5]. The efficiency CHP H. POWER CONSUMPTION
system is calculated by equation 23: In SHB there are schedulable smart appliances and fixed
P
We + Qth appliances which consume power from EG and RES. Power
∂= (23) consumption of load and HVAC is discussed below in detail.
Qf
In the above equation symbol ∂ is the total system efficiency, I. LOAD CONSUMPTION MODEL
We is the valuable electric output, Qth is the total thermal We suppose that SHB has different appliances loads L(t)
output and Qf is the fuel energy input. A CHP system at 24 hour duration which arrive over the time slots with
achieve about 60 to 80 percent efficiency. temporal variability and uncertainty. Each L(t) has power
The important fuel source for CHP is biomass, mostly pro- rating ς for a duration . In this work, the appliances are
duced from the forests. From the decomposition of organic classified in two kinds: smart schedulable appliances (SSA).
matter the biogas is obtained in the absence of oxygen in a SSA can operate themselves such as washing machines, dish
controlled tanks or landfills. CH4 is the main components washer, air conditioners, refrigerator and manually-operated
of biogas . CH4 concentration generally ranges from 30% [12] while fixed loads does not incur in bill or PAR reduction.
to 65%. CH4 material of biogas can be used for thermal and Consider that the SH has two major sets of appliances
electrical power generation as a transport fuel. Now, biogas i.e. Lsch and Lnsch , where Lsch is the set of schedulable
is an active means for producing RE, so it in turn plays i.e., can be shifted to operate in low peak slots, Lsch (t) =
a significant part in energy production and is eco-friendly. {a1 , a2 , a3 ...am } and Lnsch is the set of non-schedulable
There are numerous means for gaining biogas in the urban appliances i.e., immediately operate on the time and prefer-
areas like waste of Industries, pruning waste, solid waste, ence set by the consumer, Lnsch (t) = {b1 , b2 , b3 ...bn } over a
municipal and industrial wastewater. The electrical power scheduling duration of t ={1, 2, 3, ..., 24}. The 24 hours en-
generation through biogas is defined in equation (24). This ergy consumption of schedulable and non-schedulable load
equation is associated to the biogas conversion to valuable are given by equations (27) and (28) respectively.
energy of heat and electricity [5].
L ∗ Qbr (t) ∗ ∂ 24  X
m 
Ebio (t) = (24) Lsch (t) =
X
Lsch t, m ∈ M = {Lsch t1, m ∈ M +
γe
t=1 M =1
Where Ebio (t) is the available electrical power (kWh) at slot
Lsch t2, m ∈ M + . . . + Lsch t24, m ∈ M }
t, L is lower biogas calorific value, Qbr is biogas recovery
(27)
and availability, ∂ is efficiency and γe is recovery efficiency.
The total power P of micro-CHP generator is calculated using
24  X
m 
equation (25) as in [5]. Where CF denote capacity factor; is X
Lnsch (t) = Lsch t, m ∈ M = {Lnsch t1, m ∈
the plant accessibility factor which is taken in [80-90]% and
t=1 M =1
t is duration in hours.
M + Lnsch t2, m ∈ M + . . . + Lnsch t24, m ∈ M }
Ebio (t) (28)
P (t) = (25)
t ∗ CF
J. HVAC CONSUMPTION MODEL
G. EV ON BOARD PV HVAC is other shift-able appliance for heating, cooling and
ventilation in a building. HVAC uses almost 50% of total
EV On board PV generated energy is used for driving tasks
consumption [25]. The HVAC generally, has two mode of
and remaining is stored in electric vehicle storage system
operations i.e., heating and air cooling. In this work, we fo-
(EVSS). We assume that the harvested amount of energy
cused on cooling in summer and heating in winter. According
from the PV source installed at EV is firstly used for driving
to [25], the indoor temperature dynamics caused by an HVAC
loads and motor [13], while, the remaining energy is stored
system could be obtained as follows:
in EVSS. The EV on-board PV harvested amount of energy
η
is gt , such that: Tt+1 = Tt + (1 − )(Ttout + e et ), ∀ t (29)
Ac
t
Where, Tt the indoor temperature and Ttout denote outdoor
X
pt = min lmτ 1τ (∆ω , τ ), gt (26)
τ =0 temperature, η e is the thermal conversion efficiency, and Ac
Pt in kW/◦ F is the overall thermal conductivity, moreover, ε =
Where, the term τ =0 lmτ 1τ (∆ω , τ ) represents the total
e−τ /ω where ω is the time constant of system.
EV load arriving at time slot t [13]. The 1τ (∆ω , τ ) indicate
whether the load lmτ is being served or not at time slot τ . The K. STORAGE MODEL
remaining energy, if any, is stored into EVSS and drawn from
1) BSS Model
it only in peak hours and if the home BSS is not sufficient.
BSS is used to stores the remaining amount of RE to serve
as backup in peak hours. It stores the energy by satisfying

VOLUME 4, 2016 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An optimal power usage scheduling in smart grid

the overcharging and depth of charging constraints. Energy M. CARBON EMISSION

charged in the BSS at time slot t is described by equation In this work, carbon emission is calculated using equation
(30) as in [32]. The electricity discharged, the electricity (37) as in [32] and [14]. Where avg(EP (t)) denotes the
charged and the self-discharging rate is also considered. The average cost of electricity per month, while η illustrate the
discharging and charging of BSS would gain or lose electrical electricity price per kWh equal to 0.20 dollars and γ repre-
energy, so turn-around the BSS efficiency is depicted as: sents the emission factor of electricity equal to 1.37, while m
k.EP dch (t) denotes number of months in one year.
BS (t) = BS (t − 1) + k.δ BS .EP ch (t) − avg(EP (t))
δ BS CO2 = (37)
(30) η.γ.m
Where, BS denote the stored energy (Ah) at time t, k is time
slot duration (hour), δ BS is the efficiency of BSS, EP ch is N. UC MODEL
the electric power (kW) provided to BSS from RES at time In this work, we compute the UC in terms of delay, thermal,
t and EP dch is the electric power (kW) provided to the load air quality and visual comfort, each details and mathematical
from BSS at time t. The battery charging and discharging formulation is given below.
constraints are given below.
EP ch (t) ≤ EP ch (31) 1) Delay comfort
UB
This comfort is related to serving time of each appliance.
EP dch (t) ≤ EP dch
LB (32) Delay comfort is calculated using equation (38) as in [32].
BSS(t) ≤ ES ch (33) Where unsch(t) denotes the time of serving in unscheduled
UB
method while sch(t) denotes the time of serving in scheduled
In order to keep the BSS in good condition and avoid deep method. Delay and electricity cost are both related to UC.
discharging or overcharging, discharge and charge rate of P
|unsch (t ) − sch (t ) |
electrical energy, and energy stored in BSS should not cross Dcomf ort = P (38)
the limits approved by the company. (sch (t ))

2) Visual comfort
2) EVSS mobile backup Visual comfort is related to the number of lights and waiting
Depending on the EV total load and total source conditions, time to serve. It will be adjusted to user preference while,
the EVSS can be charged from PV, WE system and external indoor luminous intensity is defined as in [14].
power grid and its on-board PV [13]. EVSS can be used as a N e.Le (t) .f s.ϑ.M
mobile backup, our algorithm will check for the availability Vcomf ort (t) , (39)
A
of EV at home and the battery charge level for serving as
mobile backup in peak hours to shift load. Let EV availability Where Ne denotes the number of lightening devices,
is indicated as: Vcomf ort (t) in (39) shows the indoor illumination in A
illuminated indoor area. The room luminous can be adjusted
Ev (a) = {1 if availabe at home, else 0} (34) by energy consumption level of individual lighting devices
Le (t), that have their respective source flux value fs , ϑ
Ev (s) = {1 if charging, 0 idle and − 1 if discharging}
utilization factor and M maintenance factor. The user visual
(35) Lights
comfort Vc and delay APW have an inverse relation, this
The EV battery charging level at different hours after the t
relationship can be mathematically presented as in [30]:
driving tasks and status of availability is shown in Figure 12
and charge level in Figure 13. 1
Vc ∝ Lights
(40)
APW t

L. PAR Lights Lights

APW t
= (APDh
− APSLights
h
) (41)
PAR is the ratio of peak load consumed in a time slot t and the
Lights
average of total load used over the scheduling hours duration Where APD h
is the user preference of lights and
Lights
i.e., from t = 1 to T where T = 24 hours. PAR describes the APSh is the scheduled hours for lights.
power consumption activities of the user and the operation
EG peak hours and have a direct relationship with the user 3) Indoor temperature comfort
PARs. So, it is in favor for both the electric utility and user to The temperature of adjustable HVAC system, their temper-
mitigate PAR so that energy supply and user demand balance ature is varied in a definite range which is based on user
can be sustained. For single user, it is considered as in [34]. preferences. The indoor temperature can be adjusted by
max(Ltotal (t)) varying the energy consumption of temperature adjustable
P AR = 1
PT (36) HVAC equipment. The indoor temperature is varied within
total (t)
T t=1 L the range of 20°C∼25°C [14], for a single person in the smart
home, to feel comfortable. The thermal comfort is calculated

10 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An Optimal power usage scheduling in smart grid

by equation (43). The user thermal comfort and waiting time

HV AC
APW t
have an inverse relation, this relationship can be
mathematically expressed as follows:
1
Tc ∝ HV AC
(42)
APW t

HV AC HV AC
APW t
= (APDh
− APSHV
h
AC
) (43)
HV AC
Where Tc is the thermal comfort, APD h
is user pref-
HV AC
erence of HVAC and APSh is the scheduled hours for
HVAC.

4) Air quality
Air quality is measured in terms of carbon emission in the
environment. The concentration of indoor carbon can be
varied through adaptive ventilation system in the SHB. Thus,
based on user preferences, indoor good air quality can be kept Figure 3: Real time pricing signal
by ventilating fresh air into the indoor area. Mathematically,
the parameters of the indoor carbon concentration can be
stated as in [14]:
Fair (ζ out −ζ t ) + ζ in
ζ t+1 = ζ t + (44)
V
This equation (44) shows that the carbon concentration ζt+1
in an indoor zone of volume V that can be adjusted by vary-
ing the amount of fresh air Fair in the zone with respect to its
accumulation value of the CO2 concentration ζt depending
upon the outdoor carbon concentration ζout and the indoor
carbon generation ζin . The fresh air cooling and heating can
be adjusted by complementing the equation (29) when the
cooling mode is required. The desired range of indoor fresh
air is taken in terms of carbon concentration which ranges
between 740ppm∼780ppm [14].

V. RESULTS AND DISCUSSION

Figure 4: Forecasted temperature
The results and simulations of proposed LSEMC algorithm
are presented in this section. In our system model, the
incorporation of RES, BSS, EVSS and proposed HGPDO
contributes from the wind and CHP. Figure 8, 9 and 10 shows
algorithm performance is evaluated and discussed in three
the estimated power generation of solar, CHP, wind and the
cases. In case one, only EG (excluding RES and BSS),
remaining RE after BSS charging respectively . Furthermore,
while, the case two is EG with RES, and the case three is
figure 11 shows charging level of BSS along with EVSS.
EG combine with both BSS and RES. For simulations, we
used the MATLAB simulation tool. To discuss the suggested
A. CASE 1: ONLY EG
LSEMC, we assumed a smart home having six shiftable
appliances and EG, RES and BSS as a source. While BSS In this scenario, the smart home building is only using the
and EVSS is only used at peak price hours for load shifting. external grid power for scheduled and unscheduled loads.
The smart grid signals; RTP [32], forecasted temperature [4], We will discuss electricity bill cost (EBC), PAR and carbon
wind speed [4], biogas output [5] and solar irradiance [34] emission.
taken in the suggested LSEMC are showed in Figures 3, 4,
5, 6 and 7 respectively. The electricity generation by solar 1) Electricity bill cost
system primarily depends on ambient temperature and solar Figure 14 demonstrates the EBC of unscheduled load and
irradiance, while, wind and CHP depends on wind speed and scheduled load without BSS and RESs. In unscheduled, the
biogas recovery efficiency. We have assumed 90% of total maximum electricity cost is 1091 cents in slot hour 9 and
RES in all available time slots of the scheduled time. As well, in hour slot 16, the minimum cost is 165.81 cents. In case
30% of RES in each time slot is consumed for BSS charging, of BPSO, the maximum electricity cost is 727.41 cents in
of which the 20% contributes from the solar, while 10% the hour slot 9 and minimum cost is at hour slot 19. In

VOLUME 4, 2016 11

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An optimal power usage scheduling in smart grid

Figure 5: Forecasted wind speed Figure 8: Solar power generation

Figure 6: Biogas Qbr availability Figure 9: CHP power generation

Figure 7: Solar irradiance Figure 10: Wind power generation

12 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An Optimal power usage scheduling in smart grid

case of WDO, the maximum electricity cost is 699.70 cents

in the hour slot 9. In HGPDO algorithm, maximum cost
in the hour slot 7 is 386.64 cents, whereas in BFO, it is
488.83 cents in time hour slot 9. It is 727.41 cents in the
time slot 9 in GA based scheduling method. The HGPDO
algorithm performance is better in terms of EBC reduction
when compared with the other exploratory algorithms. The
average electricity bill over 24 hours in unscheduled, GA,
WDO, BPSO and BFO are 337.34, 224.89, 170.92, 324.17,
160.07 cents, respectively. While with our proposed algo-
rithm, it is 138.82 cents. The cumulative, electricity bill cost
illustrates that GA, BPSO, WDO, BFO and HGPDO reduce
the electricity bill cost by 33.33%, 3.90%, 49.33%, 52.54%
and 58.69%, respectively. Nevertheless, in this case, the best
cost minimization is achieved with the proposed scheduling
algorithm. The average cost over 24 hours of all algorithms
Figure 11: BSS and EVSS Charging
as compared with the proposed model is shown in Table 3.

Table 3: Case 1 cost comparison

Algorithm Average cost (cents) Difference (average cents) Reduction (%)
Unscheduled 337.34 -- --
GA 224.89 112.45 33.33%
BPSO 324.17 13.17 3.90%
WDO 170.92 166.42 49.33%
BFO 160.07 177.27 52.54%
HGPDO 138.82 198.52 58.69%

2) PAR
Figure 15 shows the PAR in Case 1, when SHB is only
utilizing energy from EG. In unscheduled it is 3.302, while
in GA, PSO, WDO and HGPDO it is 3.037, 2.964, 2.836,
2.654 and 1.559 respectively. Results demonstrate that GA,
PSO, WDO, BFO and HGPDO reduce the PAR by 8.025%,
10.23%, 14.11%, 19.62% and 52.78% respectively. Although
the PSO and WDO shift load to low peak price hours which
create new peaks and disturb the utility system. A penalty
Figure 12: EV availability at home
is required to be imposed for preventing the system from
such disturbance. Table 4 shows the PAR comparison of all
algorithms with the proposed algorithm for the Case 1.

Table 4: Case 1 PAR comparison

Algorithm Total PAR Difference (PAR) Reduction (%)
Unscheduled 3.302 -- --
GA 3.037 0.265 8.025%
BPSO 2.964 0.338 10.23%
WDO 2.836 0.466 14.11%
BFO 2.654 0.648 19.62%
HGPDO 1.559 1.743 52.78%

3) Carbon emission
Carbon emission in Case 1 is shown in Figure 16. Where
the desired carbon (D) is 760ppm. In case of unscheduled
it is 1014 ppm, which shows carbon emission is far from the
desired range. While in GA, BPSO, WDO, BFO and HGPDO
Figure 13: EVSS Charge level at different slots it is 703.7 ppm, 535.5 ppm, 393.1 ppm 534.8 ppm and 72.4
ppm respectively. The reduction in carbon emission in this

VOLUME 4, 2016 13

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An optimal power usage scheduling in smart grid

Figure 14: Case 1 electricity bill cost Figure 16: Case 1 carbon emission in (ppm)

B. CASE 2: EG WITH RES INTEGRATION

In this scenario, the smart building is using the external grid

power along with RES for scheduled and unscheduled task.
We will discuss EBC, PAR and carbon emission.

1) Electricity bill cost

Figure 17 demonstrates the EBC of unscheduled load and

scheduled load utilizing power from EG and RES. In un-
scheduled, the maximum electricity cost is 722.35 cents in
slot hour 9 and in slot hour 16, the minimum cost is 109.10
Figure 15: Case 1 PAR cents. In BPSO, the maximum cost is 581.13 cents in the
hour slot 9 and minimum cost is at hour slot 19. In case of
WDO, the maximum electricity cost is 496.55 cents in the
case for GA, BPSO, WDO, BFO and HGPDO is 7.40%, slot hour 9. In HGPDO algorithm, cost is 322.24 cents in the
29.53%, 48.27%, 29.63% and 72.40% respectively. In this slot hour 8, while in BFO, it is 484.37 cents at slot hour 9. In
case, WDO and HGPDO show good results in case of carbon GA centered scheduling method, cost is 666.79 cents in slot
emission reduction. Table 5 illustrate the comparisons of user hour 9. The performance of the HGPDO algorithm in terms
desired, unscheduled carbon emission and emission by all of EBC reduction is better when compared with the other
scheduling algorithms. discussed exploratory algorithms. The average electricity bill
over 24 hours in unscheduled method, GA, BPSO, WDO
and BFO are 225.91, 208.53, 216.23, 187.45 and 168.82
cents, respectively. While with our proposed algorithm it is
Table 5: Case 1 carbon emission comparison 118.48 cents. In cumulative, electricity bill cost illustrate that
GA, BPSO, WDO, BFO and HGPDO reduce the electricity
Algorithm Carbon emission (ppm) Difference (ppm) Reduction (%) bill cost by 7.69%, 4.21%, 17.02%, 22.27% and 47.55%,
CO2 concentration 760 -- --
Unscheduled 1014.0 -254 -33.42% (increased) respectively. Nevertheless, in this Case 2 the best cost mini-
GA 703.7 56.30 7.40% mization is achieved with the proposed scheduling algorithm.
BPSO 535.5 224.5 29.53%
WDO 393.1 336.9 48.27% The average cost over 24 hours duration comparison of all
BFO 534.8 225.2 29.63%
HGPDO 72.4 687.6 72.40%
algorithms and the proposed algorithm in case two is shown
in Table 6.

14 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An Optimal power usage scheduling in smart grid

Figure 17: Case 2 electricity bill cost Figure 18: Case 2 PAR

Table 6: Case 2 cost comparison it is 1003 ppm, which shows carbon emission is far from the
Algorithm Average cost (cents) Difference (average cents) Reduction (%) desired range. While in GA, BPSO, WDO, BFO and HGPDO
Unscheduled 225.91 – –
GA 208.53 17.38 7.69%
it is 682.1 ppm, 446.8 ppm, 360.7 ppm 594 ppm and 53.4
BPSO 216.23 9.68 4.21% ppm, respectively. The reduction of carbon emission in this
WDO 187.45 38.46 17.02%
BFO 168.82 57.09 25.27% case for GA, BPSO, WDO, BFO and HGPDO is 10.50%,
HGPDO 118.48 107.43 47.55% 41.21%, 52.60%, 21.84% and 92.90% respectively. In this
case, WDO and HGPDO shows good results in terms of
carbon emission reduction. Table 8 illustrate the comparison
of desired, unscheduled carbon emission and emission by all
2) PAR
scheduling algorithms.
Figure 18 shows the PAR in Case 2, when SHB is utilizing
energy from EG and RESs. In unscheduled, it is 2.985, while Table 8: Case 2 carbon emission comparison
in the case of GA, PSO, WDO, BFO and HGPDO it is Algorithm Carbon emission (ppm) Difference (ppm) Reduction (%)
2.803, 2.523, 2.453, 2.185 and 1.651 respectively. The results CO2 concentration 760 -- --
Unscheduled 1003 -243 -31.97%(increased)
demonstrate that GA, PSO, WDO, BFO and HGPDO reduce GA 682.1 77.9 10.50%
the PAR by 6.13%, 15.47%, 17.82%, 26.80% and 45.02% BPSO
WDO
446.8
360.7
313.2
399.8
41.21%
52.60%
respectively. BFO and our proposed algorithm reduce PAR BFO 594 166 21.84%
HGPDO 53.4 706.6 92.90%
very efficiently. Although the PSO and WDO shift load to
low peak price hours which create new peaks and disturb
the utility system. A penalty is required to be imposed for
preventing the system from such disturbance. Table 7 shows
C. CASE 3: EG WITH RES AND BSS INTEGRATION
the PAR comparisons of all algorithms with the proposed
algorithm for the Case 2. In this scenario, the smart building using the external grid
power for scheduled and unscheduled task along with RES
Table 7: Case 2 PAR comparison and BSS and EVSS. We will discuss EBC, PAR, carbon
Algorithm Total PAR Difference (PAR) Reduction (%)
emission and UC.
Unscheduled 2.985 -- --
GA 2.803 0.183 6.13% 1) Electricity bill cost
BPSO 2.523 0.462 15.47% Figure 20 demonstrates the EBC of unscheduled load and
WDO 2.453 0.532 17.82%
BFO 2.185 0.800 26.80% scheduled load, utilizing power from EG with RES and BSS.
HGPDO 1.651 1.344 45.02% In unscheduled, the maximum electricity cost is 711.82 cents
in slot hour 9 and minimum cost is 102.88 cents in slot hour
16. In BPSO, the maximum electricity cost is 503.57 cents
in the slot hour 9 and minimum cost is at slot hour 19. In
3) Carbon emission WDO, the maximum cost of electricity is 522.29 cents in the
Carbon emission in Case 2 is shown in Figure 19. Where slot hour 9. In HGPDO algorithm, cost is 423.48 cents in
the desired carbon (D) is 760ppm. In case of unscheduled, the slot hour 8, while in BFO, it is 444.97 cents in slot hour

VOLUME 4, 2016 15

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An optimal power usage scheduling in smart grid

Figure 19: Case 2 carbon emission in (ppm) Figure 20: Case 3 electricity bill cost

9. In case of GA based scheduling technique, it is 652.51

cents in the slot hour 9. The HGPDO algorithm performance
is better in terms of EBC reduction when compared with
the other discussed algorithms. The average electricity bill
over 24 hours in unscheduled, GA, BPSO, WDO and BFO
are 222.51, 203.97, 213.30, 175.05 and 156.38 cents, re-
spectively. While with our proposed algorithm it is 147.61
cents. The cumulative, electricity bill cost illustrates that
GA, BPSO, WDO, BFO and HGPDO reduces the electricity
bill cost by 9.68%, 4.13%, 21.32%, 29.72% and 33.66%,
respectively. Nevertheless, in this Case 3 the best cost min-
imization is achieved with proposed scheduling algorithm
which uniformly distribute load over low and high peaks
hours. The average cost over 24 hours duration comparison
of all algorithms and proposed algorithm in Case 3 is shown
in Table 9.
Figure 21: Case 3 PAR
Table 9: Case 3 cost comparison
Algorithm Average cost (cents) Difference (average cents) Reduction (%)
Unscheduled 222.51 -- --
GA 203.97 21.54 9.68%
versely, the PSO and HGPDO algorithms allocate the load
BPSO 213.30 9.21 4.13% equally and attain the required objective. Table 10 present
WDO 175.05 47.46 21.32%
BFO 156.38 66.13 29.72%
the PAR comparisons of all algorithms with the proposed
HGPDO 147.61 74.9 33.66% algorithm in the Case 3.

Table 10: Case 3 PAR comparison

Algorithm Total PAR Difference (PAR) Reduction(%)
2) PAR Unscheduled 2.537 -- --
Figure 21 demonstrates the PAR of user scheduled and un- GA 2.197 0.340 13.40%
BPSO 2.049 0.488 19.23%
scheduled load. In unscheduled, it is 2.537, while in the case
WDO 1.741 0.796 31.37%
of GA, PSO, WDO, BFO and HGPDO it is 2.197, 2.049, BFO 1.727 0.810 31.92%
1.741, 1.727 and 1.158 respectively. The results demonstrate HGPDO 1.158 1.379 54.35%
that the PAR is reduced as a result of proposed GA, PSO,
WDO, BFO and HGPDO algorithms by 13.40%, 19.23%,
31.37%, 31.92% and 54.35%, respectively. Nevertheless, the
HGPDO algorithm curtail the PAR significantly as compared 3) Carbon emission
with the other exploratory algorithms. The load is shifted to Carbon emission in Case 3 is shown in Figure 22. Where
off peak slots by GA and WDO and create new peaks. Con- the desired carbon (D) is 760ppm. In case of unscheduled, it

16 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An Optimal power usage scheduling in smart grid

Figure 22: Case 3 carbon emission in (ppm)

Figure 23: Waiting time of all appliances

is 968.3 ppm, which shows carbon emission is far from the Table 12: Delay comfort (minutes)
desired range. While in GA, BPSO, WDO, BFO and HGPDO
Appliances GA BPSO WDO BFO HGPDO Comfort(%) by HGPDO
it is 671.6 ppm, 541.6 ppm, 374.8 ppm 567.5 ppm and 63.5 Water heater (WH) 37 37 37 37 41 31.66%
Refrigerator 100 100 60 45 45 25.00%
ppm, respectively. The reduction in carbon emission in this HVAC 0 16 33 33 50 16.66%
case for GA, BPSO, WDO, BFO and HGPDO is 11.63%, Washing machine (WM)
EV
04
100
12
70
12
100
12
62
32
31
46.66%
48.33%
28.73%, 50.68%, 25.39% and 91.64% respectively. In this Lights 09 29 29 29 33 45.00%

case, WDO and HGPDO shows better results in terms of car-

bon emission reduction. Table 11 illustrates the comparison
2) Thermal comfort
of desired, unscheduled carbon emission and emission by all
scheduling algorithms. Thermal comfort is calculated using equations (29) and (43)
which is related to HVAC, cooling and heating (summer and
Table 11: Case 3 carbon emission comparison winter season) and user given preference. Figure 24 shows
the thermal comfort of our proposed HGPDO, GA, WDO,
Algorithm Carbon emission (ppm) Difference (ppm) Reduction (%)
CO2 concentration 760 -- -- PSO and BFO algorithms. BPSO and WDO returns best
Unscheduled 968.3 -208.30 -27.40% (increased)
GA 671.6 88.40 11.63%
values of TC. Also, it schedules the HVAC load on urgent
BPSO 541.6 218.40 28.73% basis with zero delay.
WDO 374.8 385.20 50.68%
BFO 567.5 193.00 25.39%
HGPDO 63.5 696.5 91.64%
3) Visual comfort
Visual comfort is calculated using equations (39) and (41)
which is related to number of lights, luminous intensity and
user given preference. Figure 24 shows the visual comfort
D. USER COMFORT of our proposed HGPDO, GA, WDO, PSO and BFO. WDO
1) Delay and cost comfort returns the best values of VC because it schedules lights on
UC is related to both EBC and scheduling wait time. Comfort urgent basis with zero delay.
is considered using equation (38) in terms of scheduling wait
time in this work. Scheduling wait time is that the consumer 4) Air quality comfort
must wait to turn on their appliances. Consumer will control Air quality comfort is calculated using equations (37) and
their home appliances agreeing to scheduling pattern pro- (44). The AQC is related to carbon emission, electricity con-
vided by controller for reduction in EBC. A consumer who sumption and desired air freshness factor. Figure 25 shows
desires to reduce the cost, he will compromise on comfort the air quality comfort of our proposed HGPDO, GA, WDO,
level. Figure 23 shows the average scheduling wait time of PSO and BFO. The proposed algorithm reduces the carbon
all home appliances scheduled by GA, HGPDO, WDO, PSO emission, which results in less concentration of CO2 in air,
and BFO algorithms respectively. GA has no wait time in hence, the optimal AQC is achieved.
serving HVAC load. Our proposed algorithm scheduling wait
time of all appliances is less than one hour. Table 12 displays VI. CONCLUSION AND FUTURE WORK
the scheduling time of the home appliances in minutes and The authors have proposed an efficient load scheduling and
respective improvement in case of HGPDO algorithm. energy management controller for smart home building to

VOLUME 4, 2016 17

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An optimal power usage scheduling in smart grid

gration of RES and BSS reduces the electricity bill, PAR

and CO2 in Case 1, by 58.69%, 52.78% and 72.40%, in
Case 2, by 47.55%, 45.02% and 92.90% and in Case 3, by
33.6%, 54.35% and 91.64%, respectively as compared with
unscheduled. Moreover, the user comfort by our proposed
HGPDO algorithm in terms of delay, thermal, air quality
and visual improved by 35.55%, 16.66%, 91.64% and 45%,
respectively.
This work is based on domestic smart home load in residen-
tial sector. It can be applied to a commercial and industrial
sector, and can also be considered with a large number of
appliances and onsite renewable energy generation.
In future, we will work on real time algorithms and will
evaluate the performance of our proposed system for the
same scenarios.

References
Figure 24: Thermal and visual comfort in minutes [1] International Energy Agency Accessed: Mar. 21, 2021. [Online].
Available:https://www.researchandmarkets.com/issues/electricity-
demand.
[2] World Nuclear Association. Accessed: Mar. 21, 2021. [Online].
Available:https://www.world-nuclear.org/nuclear-essentials/where-
does-ourelectricity-come-from.aspx
[3] International Energy Agency Accessed: Mar. 21, 2021. [Online].
Available:https://www.homeselfe.com/average-cost-electric-bills-rising-
across-u-s/
[4] Time and date forcaste, Hour-by-Hour Forecast for Is-
lamabad, Pakistan: : Mar. 21, 2021. [Online]. Avail-
able:https://www.timeanddate.com/weather/pakistan/islamabad/hourly
[5] A. B. Escandón,J. M. O. Ruiz, J. D. Curillo and E. F. Zalamea-León.”
Assessment of Power Generation Using Biogas from Landfills in an
Equatorial Tropical Context”, Sustainability 2020, 12, 2669.
[6] K. Aurangzeb, S. Aslam, H. Herodotou, M. Alhussein and S. I. Haider,
"Towards Electricity Cost Alleviation by Integrating RERs in a Smart
Community: A Case Study," 2019 23rd International Conference Electron-
ics, 2019, pp. 1-6.
[7] X. Jiang and C. Xiao, “Household energy demand management strategy
based on operating power by genetic algorithm,” IEEE Access, vol. 7, pp.
9641496423, 2019.
[8] Judge, Malik Ali, Awais Manzoor, Carsten Maple, Joel JPC Rodrigues,
and Saif ul Islam. "Price-based demand response for household load
management with interval uncertainty." Energy Reports (2021).
[9] K. Ullah, S. Ali, T. A. Khan, I. Khan, S. Jan, I. A. Shah, and G. Hafeez,
Figure 25: Air quality comfort in (ppm) “An Optimal Energy Optimization Strategy for Smart Grid Integrated with
Renewable Energy Sources and Demand Response Programs,” Energies,
vol. 13, no. 21, p. 5718, Nov. 2020.
[10] T. Li and M. Dong, “Real-Time Residential-Side Joint Energy Storage
reduce the electricity bill, PAR, carbon emission and improve Management and Load Scheduling With Renewable Integration,” in IEEE
Transactions on Smart Grid, vol. 9, no. 1, pp. 283-298, Jan. 2018.
UC in terms of visual, thermal, air quality and delay. This [11] T. Sattarpour, D. Nazarpour, and S. Golshannavaz, “A multi-objective
work considers a smart building utilizing power from EG, HEM strategy for smart home energy scheduling: A collaborative ap-
BSS and RES i.e., solar, thermal and wind power to shift load proach to support microgrid operation,” Sustain. Cities Soc., vol. 37, pp.
2633, Feb. 2018.
to peak hours and achieve the objectives. The huge HVAC
[12] A. Nawaz et al., "An Intelligent Integrated Approach for Efficient Demand
load is scheduled with other different shiftable and smart Side Management With Forecaster and Advanced Metering Infrastructure
appliances to reduce load. Each appliance in the smart home Frameworks in Smart Grid," in IEEE Access, vol. 8, pp. 132551-132581,
2020.
is scheduled using GA, BPSO, WDO, BFO and the proposed
[13] A. Ahmad and J. Y. Khan, “Real-Time Load Scheduling and Storage
optimization technique, HGPDO. The proposed technique Management for Solar Powered Network Connected EVs, in IEEE Trans-
helps to find the most optimum schedule of each home ap- actions on Sustainable Energy,” vol. 11, no. 3, pp. 1220-1235, July 2020.
pliance considering system constraints. The performance of [14] A. Ahmad, J. Y. Khan, “Real-Time Load Scheduling, Energy Storage
Control and Comfort Management for Grid-Connected Solar Integrated
the proposed scheme and heuristic algorithms are evaluated Smart Buildings, Applied Energy,” Volume 259, 2020, 114208, ISSN
using real time pricing scheme via MATLAB simulations. 0306-2619.
Moreover, we compared the results of the proposed method [15] A. Ahmad and J. Y. Khan, “Roof-Top Stand-Alone PV Micro-Grid: A Joint
Real-Time BES Management, Load Scheduling and Energy Procurement
with the GA, PSO, WDO and BFO to check its efficiency. From a Peaker Generator,” in IEEE Transactions on Smart Grid, vol. 10,
Results demonstrate that the proposed algorithm and inte- no. 4, pp. 3895-3909, July 2019.

18 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An Optimal power usage scheduling in smart grid

[16] Hafeez, Ghulam, Zahid Wadud, Imran Ullah Khan, Imran Khan, Zeeshan [35] Hafeez, Ghulam, Noor Islam, Ammar Ali, Salman Ahmad, Muhammad
Shafiq, Muhammad Usman, and Mohammad Usman Ali Khan. “Efficient Usman, and Khurram Saleem Alimgeer. "A Modular Framework for
Energy Management of IoT-Enabled Smart Homes Under Price-Based Optimal Load Scheduling under Price-Based Demand Response Scheme
Demand Response Program in Smart Grid.” Sensors 20, no. 11 (2020): in Smart Grid.” Processes 7, no. 8 (2019): 499.
3155. [36] C. Bharathi, D. Rekha, & V. Vijayakumar, “Genetic Algorithm Based
[17] Hafeez, Ghulam, Khurram Saleem Alimgeer, Zahid Wadud, Imran Khan, Demand Side Management for Smart Grid.” Wireless Pers Commun 93,
Muhammad Usman, Abdul Baseer Qazi, and Farrukh Aslam Khan. “An p. 481–502, 2017.
Innovative Optimization Strategy for Efficient Energy Management with
Day-ahead Demand Response Signal and Energy Consumption Forecast-
ing in Smart Grid using Artificial Neural Network.” IEEE Access 8 (2020):
84415-84433.
[18] S. Umetani, Y. Fukushima, H. Morita, “A linear programming based
heuristic algorithm for charge and discharge scheduling of electric vehicles ATEEQ UR REHMAN is pursuing M.Sc. degree
in a building energy management system,” Omega, Volume 67, 2017, in Computer System Engineering from the Univer-
Pages 115-122, ISSN 0305-0483 sity of Engineering and Technology Peshawar. He
[19] A. R. Kalair, N. Abas, Q. U. Hasan, M. Seyedmahmoudian, N. Khan, has authored or co-authored over in peer-reviewed
“Demand side management in hybrid rooftop photovoltaic integrated research papers in reputed international journals
smart nano grid, Journal of Cleaner Production,” Volume 258, 2020, and conferences. His research interests include
120747,ISSN 0959-6526. optimization, planning, energy management, and
[20] K. Aurangzeb, M. Alhussein, K. Javaid and S. I. Haider, "A Pyramid- machine learning applications in smart grids /mi-
CNN Based Deep Learning Model for Power Load Forecasting of Similar- crogrids, etc
Profile Energy Customers Based on Clustering," in IEEE Access, vol. 9,
pp. 14992-15003, 2021.
[21] K. G. D. Santo, S. G. D. Santo, R. M. Monaro, M. A. Saidel, “Active
demand side management for households in smart grids using optimization
and artificial intelligence,” Measurement, Volume 115, 2018, Pages 152-
161, ISSN 0263-2241.
[22] B. Celik, R. Roche, S. Suryanarayanan, D. Bouquain, A. Miraoui, “Electric ZAHID WADUD recently completed his Ph.D.
energy management in residential areas through coordination of multiple from the Capital University of Science and Tech-
smart homes,” Renewable and Sustainable Energy Reviews, Volume 80, nology, Islamabad Pakistan with the thesis enti-
2017, Pages 260-275, ISSN 1364-0321. tled, “Energy balancing with sink mobility in the
[23] M. Killian, M. Zauner, and M. Kozek, “Comprehensive smart home energy design of underwater routing protocols”. Previ-
management system using mixed-integer quadratic-programming,” Appl. ously, he has completed BSc and Master’s degrees
Energy, vol. 222, pp. 662672, Jul. 2018. in Electrical Engineering from University of En-
[24] L. Yu, T. Jiang and Y. Zou, "Distributed Real-Time Energy Management gineering and Technology Peshawar Pakistan in
in Data Center Microgrids," in IEEE Transactions on Smart Grid,” vol. 9, 1999 and 2003 respectively. Currently he is work-
no. 4, pp. 3748-3762, July 2018.
ing as Assistant Professor in the Department of
[25] L. Yu, T. Jiang and Y. Zou, "Online Energy Management for a Sustainable
Computer Systems Engineering, University of Engineering and Technology
Smart Home With an HVAC Load and Random Occupancy," in IEEE
Peshawar, Pakistan. His research interests include wireless sensor network,
Transactions on Smart Grid, vol. 10, no. 2, pp. 1646-1659, March 2019.
[26] M. A. Hannan, M. G. M. Abdolrasol, M. Faisal, P. J. Ker, R. A. Begum energy efficient networks and subsystems, mathematical modeling of wire-
and A. Hussain, "Binary Particle Swarm Optimization for Scheduling MG less channels, embedded systems and sensors interface. He published over
Integrated Virtual Power Plant Toward Energy Saving," in IEEE Access, dozen state-of-the-art publications in the renowned International journals.
vol. 7, pp. 107937-107951, 2019.
[27] B. Celik, R. Roche, S. Suryanarayanan, D. Bouquain, A. Miraoui, Electric
energy management in residential areas through coordination of multiple
smart homes, Renewable and Sustainable Energy Reviews, Volume 80,
2017, Pages 260-275, ISSN 1364-0321. RAJVIKRAM MADURAI ELAVARASAN re-
[28] S. Williams, M. Short, Electricity demand forecasting for decentralized ceived the B.E. degree in electrical and electron-
energy management, Energy and Built Environment, Volume 1, Issue 2, ics engineering from Anna University, Chennai,
2020, Pages 178-186, ISSN 2666-1233.
India, and the M.E. degree in power system engi-
[29] K. Aurangzeb, S. Aslam, Haider, SI, et al. "Energy forecasting using
neering from the Thiagarajar College of Engineer-
multiheaded convolutional neural networks in efficient renewable energy
resources equipped with energy storage system," Trans Emerging Tel ing, Madurai. He worked as an Associate Tech-
Tech. 2019;e3837. nical Operations with the IBM Global Technol-
[30] A. Khalid, N. Javaid, M. Guizani, M. Alhussein, K. Aurangzeb and M. ogy Services Division. He worked as an Assistant
Ilahi, "Towards Dynamic Coordination Among Home Appliances Using Professor with the Department of Electrical and
Multi-Objective Energy Optimization for Demand Side Management in Electronics Engineering, Sri Venkateswara Col-
Smart Buildings," in IEEE Access, vol. 6, pp. 19509-19529, 2018. lege of Engineering, Chennai. He currently works as a Design Engineer
[31] A. Ghassemi, P. Goudarzi , M. R. Mirsarraf, and T. A Gulliver, “A Stochas- with the Electrical and Automotive Parts Manufacturing Unit, AA Industries,
tic Approach to Energy Cost Minimization in Smart-Grid-Enabled Data Chennai. He also works as a Visiting Scholar with the Clean and Resilient
Center Network”, journal of Computer Networks and Communications, Energy Systems (CARES) Laboratory, Texas AM University,Galveston, TX,
vol. 2019, Article ID 4390917, 11 pages, 2019. USA. He has published papers in international journals, and international
[32] A. Imran et al., "Heuristic-Based Programable Controller for Efficient En- and national conferences. His research interests include solar PV cooling
ergy Management Under Renewable Energy Sources and Energy Storage techniques, renewable energy and smart grids, wind energy research, power
System in Smart Grid," in IEEE Access, vol. 8, pp. 139587-139608, 2020. system operation and control, articial intelligence, control techniques, and
[33] C. Wenge, R. Pietracho, S. Balischewski, B. Arendarski, P. Lombardi, demand-side management. He received the Gold Medal for his master’s
P. Komarnicki, and L. Kasprzyk, “Multi Usage Applications of Li-Ion
degree. He is a recognized Reviewer in reputed journals, namely the
Battery Storage in a Large Photovoltaic Plant: A Practical Experience,”
IEEE SYSTEMS, IEEE ACCESS, the IEEE Communications Magazine,
Energies, vol. 13, no. 18, p. 4590, Sep. 2020.
[34] Ali, Sajjad, Imran Khan, Sadaqat Jan, and Ghulam Hafeez. "An Optimiza-
International Transactions on Electrical Energy Systems (Wiley), Energy
tion Based Power Usage Scheduling Strategy Using Photovoltaic-Battery Sources, Part A: Recovery, Utilization and Environmental Effects (Taylor
System for Demand-Side Management in Smart Grid." Energies 14, no. 8 and Francis), Scientific Reports (Springer Nature), Chemical Engineering
(2021): 2201. Journal (Elsevier), and CFD Letters and Biotech (Springer).

VOLUME 4, 2016 19

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3087321, IEEE Access

Ateeq Ur Rehman et al.: An optimal power usage scheduling in smart grid

GHULAM HAFEEZ completed his B.Sc. in Elec- HASSAN HAES ALHELOU (Senior Member,
trical Engineering from University of Engineering IEEE) received the B.Sc. degree (Hons.) from
and Technology Peshawar, Pakistan and MS and Tishreen University, Latakia, Syria, in 2011, the
PhD degree in Electrical Engineering from COM- M.Sc. degree (Hons.) from the Isfahan University
SATS University Islamabad, Islamabad, Pakistan. of Technology (IUT), Isfahan, Iran, in 2016, all
He is lifetime charted engineer from Pakistan in electrical power engineering, power systems,
Engineering Council. He is working as a Man- where he is currently pursuing the Ph.D. degree.
ager University-Industry Linkages/Research Op- He is also a Faculty Member with Tishreen Uni-
erations Development in the Directorate of ORIC, versity. He is included in the 2018 Publons’s list
University of Engineering and Technology, Mar- of the top 1% best reviewer and researchers in
dan. Prior to this, he was Lecturer in Department of Electrical Engineering, the field of engineering in the world. He has published more than 30
University of Engineering and Technology, Mardan. He also worked as research papers in the high-quality peer-reviewed journals and international
a Lecturer University of Wah, Wah Cantt, Pakistan. He has also worked conferences. He has also performed more than 160 reviews for high-
as a Research Associate in COMSATS University Islamabad, Islamabad, prestigious journals, including the IEEE TRANSACTIONS ON INDUS-
Pakistan, where his research focus was Computational Intelligence, Forecast TRIAL INFORMATICS, the IEEE TRANSACTIONS ON INDUSTRIAL
Process, Energy Management, Operation of Electricity Market, and Electric ELECTRONICS, Energy Conversion and Management, Applied Energy,
Vehicles in Smart Power Grids. His industrial experience includes work- and the International Journal of Electrical Power and Energy Systems. He
ing for Alcatel-Lucent and PTCL as Optimization Engineer in Islamabad, has participated in more than 15 industrial projects. His major research
Pakistan. He has authored or co-authored in peer-reviewed research papers interests include power systems, power system dynamics, power system op-
in reputed international journals and conferences. His research interests eration and control, dynamic state estimation, frequency control, smart grids,
include: Sustainable and Smart Energy, Cities and Societies, Smart Grids; micro-grids, demand response, and load shedding. He was a recipient of the
Applications of Deep Learning and Blockchain in Smart Power Grids; and Outstanding Reviewer Award from many journals, such as Energy Conver-
Stochastic Techniques for Power Usage Optimization in Smart Power Grids sion and Management (ECM), ISA Transactions, and Applied Energy. He
etc. was also a recipient of the Best Young Researcher in the Arab Student Forum
Creative among 61 researchers from 16 countries at Alexandria University,
Egypt, in 2011.

IMRAN KHAN (Senior Member, IEEE) received

the B.Sc. degree in Electrical Engineering from
N.W.F.P. University of Engineering and Technol-
ogy, Peshawar, Pakistan in 2003 and M.Sc. degree
in telecommunication engineering from the Asian
Institute of Technology, Thailand, in 2007. He
did Ph.D. degree at the Telecommunications FOS,
School of Engineering and Technology, Asian In-
stitute of Technology, Thailand, in 2010. Currently
he is working as professor in Electrical Engi-
neering Department, University of Engineering Technology, Mardan. His
research interests include performance analysis of Wireless Communication
Systems, OFDM, OFDMA, MIMO, Cooperative Networks, Cognitive Radio
Systems, and Energy Management in the Smart Grid.

ZEESHAN SHAFIQ received the bachelor’s,

master’s, and Ph.D. degrees in electrical engineer-
ing from the University of Engineering and Tech-
nology at Peshawar, in 2009, 2012, and 2018, re-
spectively. He was a recipient of the International
Research Support Initiative Program Scholarship
from the Higher Education Commission Pakistan
for visiting The University of Sydney, Australia.
During his stay, he was involved in unmanned
aerial vehicles at the Centre of Excellence in
Telecommunications. His research interests include vehicular ad hoc net-
works, intelligent transportation systems, and unmanned aerial vehicles. He
has also reviewed many research papers for well-reputed journals, including
IEEE ACCESS and the IEEE Technical Review Journals.

20 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/