Australian Journal of Electrical and Electronics

Engineering

ISSN: (Print) (Online) Journal homepage: www.tandfonline.com/journals/tele20

Microgrids with day-ahead energy forecasting

for efficient energy management in smart grids:
hybrid CS-RERNN

C. P. Shirley, Jagannath Pattar, P. Kavitha Rani, Sumit Saini, Jarabala Ranga,

D. Elangovan & Ch. Venkatakrishna Reddy

To cite this article: C. P. Shirley, Jagannath Pattar, P. Kavitha Rani, Sumit Saini, Jarabala Ranga,
D. Elangovan & Ch. Venkatakrishna Reddy (03 Mar 2024): Microgrids with day-ahead energy
forecasting for efficient energy management in smart grids: hybrid CS-RERNN, Australian
Journal of Electrical and Electronics Engineering, DOI: 10.1080/1448837X.2024.2312488

To link to this article: https://doi.org/10.1080/1448837X.2024.2312488

Published online: 03 Mar 2024.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=tele20
AUSTRALIAN JOURNAL OF ELECTRICAL AND ELECTRONICS ENGINEERING
https://doi.org/10.1080/1448837X.2024.2312488

Microgrids with day-ahead energy forecasting for efficient energy

management in smart grids: hybrid CS-RERNN
C. P. Shirleya, Jagannath Pattarb, P. Kavitha Ranic, Sumit Sainid, Jarabala Rangae, D. Elangovanf
and Ch. Venkatakrishna Reddyg
a
Department of Computer Science and Engineering, Karunya Institute of Technology and Sciences, Coimbatore, India; bDepartment of
Mechanical Engineering, Madanapalle Institute of Technology & Science, Madanapalle, India; cDepartment of M.Tech Computer Science
and Engineering, Sri Krishna College of Engineering and Technology, Coimbatore, India; dDepartment of Electrical Engineering, School of
Engineering & Technology, Central University of Haryana, Mahendergarh, India; eDetartment of Electrical and Electronics Engineering,
Ramachandra College of Engineering, Eluru, India; fDepartment of Computer Science and Engineering, Panimalar Engineering College,
Chennai, India; gDepartment of Electrical and Electronics Engineering, Chaitanya Bharathi institute of technology, Hyderabad, India

ABSTRACT ARTICLE HISTORY

By integrating smart grid technology with home energy management systems, households can Received 3 May 2023
monitor and optimise their energy consumption. This allows for more efficient use of energy Accepted 14 January 2024
resources, reducing waste and lowering energy bills. In this manuscript, a hybrid approach is KEYWORDS
proposed for smart grid home energy management with microgrids and day-ahead energy Efficient energy utilisation;
forecasts. The proposed control approach combines the Circle Search (CS) algorithm and energy management; energy
Recalling-Enhanced Recurrent Neural Network (RERNN). Commonly it is named as CS-RERNN forecasting; smart grid;
technique. The novelty of this paper is to optimise energy consumption and production within renewable energy; smart
microgrids, thereby contributing to the overall efficiency of the smart grid system. The home; micro gas turbine;
proposed method is used to reduce the electricity cost, peak-to-average ratio (PAR), and electric vehicle
maximising consumer comfort. Energy management is performed based on the CS algorithm.
A smart home connected to the external power grid (PG) is managed by the proposed method.
Here, load demand is predicted by using RERNN. By then, the performance of the proposed
method is implemented in MATLAB platform. The proposed method shows a high efficiency of
96%, and 22 $ of low electricity bill cost compared with other existing methods such as Particle
swarm optimisation (PSO), Cuckoo search algorithm (CSA, and Border collie Optimisation
(BCO).

1. Introduction localised energy distribution systems that can function

independently or in conjunction with the main grid.
The traditional electrical system, which depends on
These microgrids bring benefits like improved relia­
centralised fossil fuel plants located far from energy
bility, decreased transmission losses, and the combi­
consumers, suffers from power losses due to the large
nation of renewable energy sources (Gnanaprakasam
distances between power generation and users. This
et al. 2023). A technique called day-ahead energy
results in power outages, with transmission and dis­
forecasting plays a role in predictive modelling, utilis­
tribution lines being the main culprits (Gungor et al.
ing historical data and weather patterns to estimate
2012). To counter this issue, there is a need for local
energy consumption and generation for the following
power plants near consumption areas, preferably uti­
24 hours (Rajani and Naidu Kommula 2022). This
lising sustainable energy sources like microgrids
approach encourages end users to manage their energy
(MGs). MGs offer a solution to the problem of carbon
consumption and even offers incentives for peak
emissions as they reduce reliance on traditional energy
reduction through demand response (DR) schemes.
generation (Rehman et al. 2021). The energy sector is
By applying such forecasting, consumers can plan
a major contributor to carbon emissions, with
their appliance usage to reduce costs and enhance
approximately 41% generated by the energy sector,
satisfaction.
and 23% by the transport sector in terms of global
In DR, the challenge arises when numerous house­
greenhouse gases (Esther and Sathish Kumar 2016).
holds with similar lifestyles are considered. Having
The emergence of smart grids has opened doors to identical devices with the same operational hours
more efficient energy management systems capable of and power ratings in multiple homes is impractical.
meeting the increasing energy demands of modern The smart grid comes into play here, overcoming the
society (Arun Kumar et al. 2023). A vital component limitations of centralised generation (CG). Advanced
of smart grids is the integration of microgrids, Metering Infrastructure enables the bidirectional

CONTACT C. P. Shirley [email protected]

© Engineers Australia
2 C. P. SHIRLEY ET AL.

transmission of energy and information between con­ major purpose of the proposed method is to reduce
sumers and sources. The accuracy of energy demand electricity costs. The remaining of the manuscript is
and supply prediction empowers microgrid operators mentioned as below: Section 2 describes the recent
to optimise energy management strategies, leading to research work and its background, section 3 clarifies
cost reduction and improved grid stability (Gungor the structure of energy efficient model. Efficient
et al. 2011). However, scaling up this framework to energy management on smart grids based CS-
accommodate a large number of microgrids and elec­ RERNN is illustrated in Section 4, Section 5 clarifies
tric vehicles poses challenges. The complexity of coor­ the results and discussion and Section 6 concludes the
dinating and optimising their energy management manuscript.
grows as their numbers increase. Dealing with uncer­
tainties in electric vehicle charging patterns, renewable
2. Resent research work: a brief review
energy generation, and shifting energy demands com­
plicates accurate optimisation. Implementing an Various research works have earlier presented in the
intelligent day-ahead energy management framework literatures were depends on the energy management in
could encounter regulatory and policy hurdles smart grids through the use of microgrids. Some of
(Huaman and Xiu Jun 2014). them are reviewed here.
This paper’s objective is to explore how day-ahead
energy forecasting techniques can enhance microgrid
2.1. Background of research work
efficiency within smart grids. Various forecasting
models and algorithms, including statistical and The recent research work indicates that efficient
machine learning approaches, will be discussed energy management in smart grids is a critical issue.
(Zhao et al. 2016). The challenges and opportunities The MILP formulation suffers from drawbacks like
tied to day-ahead energy forecasting in microgrids will inability to account for nonlinear effects, the require­
be examined, covering factors like data availability, ment to examine the entire time periods at once, and
accuracy, real-time adjustments, renewable energy the high-dimensionality risk of the issue. Several stra­
impact, and demand response influence. The integra­ tegies are now being given for scheduling smart appli­
tion of day-ahead energy forecasting promises benefits ances in order to lower the peak to average ratio, and
like load balancing, optimal resource scheduling, peak consumer delay time. There are several drawbacks of
demand reduction, and improved grid resilience. using genetic algorithms. The motivation for doing
Ultimately, accurate energy predictions enable micro­ this research is provided by the aforementioned
grid operators to optimise operations, cut costs, and drawbacks.
contribute to a sustainable energy future.
3. Configuration of the energy efficient Model
1.1. Objectives and contribution
This section defines the efficient energy management in
● The main contribution of this paper is to propose SG using microgrids in the day-ahead energy forecast­
a hybrid CS-RERNN approach for smart grid ing. It consists of solar energy, solar energy inverter,
home energy management with microgrids solar energy gateway, wind energy gateway, wind energy,
and day-ahead energy forecasts. static batteries, smart metre, inverter, electric vehicle,
● The proposed control approach combines the CS and home appliances (Moon and Lee 2016). Structure
algorithm and RERNN. of energy efficient model is shown in Figure 1. In this,
● The objective of this manuscript is to improve the residential sector is examined, and smart-house is
energy consumption and production within linked to the exterior grid, which has a smart metre that
microgrids, thereby increasing the overall effi­ uses the proposed CS-RERNN algorithm.
ciency of the SG system and minimise the elec­
tricity bill cost.
3.1. Energy procurement
● The proposed method is executed in MATLAB
platform, and it is compared to other existing The smart-house obtains energy from exterior PG, on-
methods. site renewable sources, like wind, CHP, and solar, as
well as scheduled, urgent, and unscheduled demands.
It uses the EV battery, if available in home, as well as
1.2. Organization
the BSS at as a backup at peak-hours.
In this paper presents a hybrid CS-RERNN for energy
management in SG using microgrids and day-ahead
3.2. External power grid
energy forecasts. The proposed control strategy com­
bines the Circle Search (CS) algorithm and RERNN. Here, the smart home purchases electricity from the
Typically, it is known as CS-RERNN method. The external grid at a predetermined cost. The local power-
 AUSTRALIAN JOURNAL OF ELECTRICAL AND ELECTRONICS ENGINEERING 3

Author Main Contribution Techniques Adopted Controller used Energy management issues
Aslam et al. Efficient energy management method Mixed integer linear Energy Management It was not practically possible for
(2020) to systematically manage the energy programming Controller (EMC) multiple homes to have the same
consumption in the residential area (MILP), Artificial neural appliances with the same length of
to lessens the PER, and electricity network (ANN) operational time (LOT) and power
cost along with user comfort ratings (kW).
maximisation.
Shuja et al. Reduce the cost through load shifting Jaya algorithm (JA) on - Cost on implementation, maintenance
(2020) techniques. a bat algorithm (BA) for their system and the generator
and candidate solution fuel was not preferred in their
updation algorithm formulation. They neglect PAR and
(CSUA) user comfort.
Ullah et al. Smart microgrid by considering a novel multi-objective genetic Microgrid central Creating different contexts also adds to
(2021) hybrid scheme of DRPS and IBT as algorithm (MOGA) controller the mathematical problem’s
the program to cater to the and29 multi-objective complexity.
uncertainty caused by solar and wind-driven
wind power sources of optimisation optimisation (MOWDO)
function with conflicting objectives.
Farinis and An easy to apply and computationally Particle swarm - The autonomous operation of the
Kanellos efficient energy management optimisation (PSO) microgrid have not been considered.
(2021) system (EMS) for microgrids of
building prosumers.
Taherian To support a retail electric provider Deep learning (DL), Auto- - It was difficult and impossible to
et al. (REP) to make the best day-ahead encoded (AE) neural determine a fixed pattern behavior
(2021) dynamic pricing decisions in network for the lifestyle of some customers.
a realistic scenario.
Vosoogh An intelligent stochastic framework for Particle Swarm Central control approach Implementing an intelligent day ahead
et al. optimal operating and managing of optimisation (PSO), energy management framework for
(2020) these systems with high penetration Genetic Algorithm (GA) networked microgrids considering
of renewables. high penetration of electric vehicles
can be complex and required
advanced technical knowledge.
Garcia- A new procedure based on MPC for the Model predictive control Predictive Plant Model of The risks can seriously affect the
Torres energy regulation among microgrids (MPC) based controller the Microgrid, MPC economic optimisation of the
et al. inside local networks with an Algorithm for the microgrids, which under unexpected
(2020) internal electrical market energy regulation conditions were forced to participate
between microgrids in disadvantageous conditions in the
intraday.

grid system supplies electricity to the smart home. Let here, EPðtÞ refers electricity price on particular
Eex ðtÞ stand in for the total amount of energy taken time slot t. The below equation can be used at
from the external PG at time t. each time slot t after considering the net RESs
and BSS:
0 � Eex ðtÞ � Eex;max (1)
X
a
a
lbill ¼ lbill ðEpv ðtÞ þ pwd ðtÞ þ pchp ðtÞÞ BSSðτÞ
m¼1
3.3. Electricity price model (4)
Here, a day-ahead pricing (DAP) and flat-price
(FP) is utilised from which day-ahead pricing is here, τ as time period in which the renewable
an advance price sent to the user thru a net metre energy sources are either inaccessible, which leads
and flat-price is an hourly-price that varies and to consume energy as battery storage system to
stays constant for the course of an hour (Aslam load.
et al. 2017). Equation 2 calculate the energy expen­
ditures for home-appliances which can be sched­
uled (lsch ) and those that cannot (lnsch ) during a 24- Problem Formulation: The main objective function
hour period. is to lessen the cost of electrcity bill.
0 1 !
X24 X
T X
a
@ P ðla ðtÞ � sa ðtÞÞ � EPðtÞÞ A
a M a
lbill ¼ min imize lbill ðtÞ ðϕðtÞ þ BSSðtÞÞ (5)
m2M m2M
t¼1 m¼1 t¼1 m¼1

(2) Based on the limits (6)-(9)

here, A ¼ fsch; nschg
� X
n
a
1 if l is ON η ¼ LðAN Þ (6)
sam2M ðtÞ ¼ (3)
0 if la is OFF A¼1
4 C. P. SHIRLEY ET AL.

Figure 1. Structure of energy efficient model.

X
N 3.4. Solar generation system
α�η�β (7)
a¼1 The amount of solar energy captured the effective­
ness of the energy transfer, and the positioning
ϕðtÞ ¼ ppv ðtÞ þ pwd ðtÞ þ pchp ðtÞ (8) and orientation of the panels all determine how
much solar energy is generated. Throughout the
course of a 24-hour period, the produced energy is
0 � BSSMIN � BSSMAX for all t 2 T (9)
typically distributed to specify loads during each
where L as working time for the appliance, AN as time slot.
particular appliance, ϕðtÞ as overall purchases of
renewable energy at time t, η as duration time of epv ðtÞ ¼ γpv � apv � ird ðtÞ � ð1 0:005Þ
working, α as time of start for smart home appli­ � ðt0 ðtÞ 25Þ for all t (10)
ance, β as time of end.
 AUSTRALIAN JOURNAL OF ELECTRICAL AND ELECTRONICS ENGINEERING 5

here, γpv as Effectiveness of Energy conversion factor l � qbr ðtÞ � @e

ebio ðtÞ ¼ (14)
for solar panels, apv as surface area of panel in m2, γe
ird ðtÞ as radiance of solar in kW/m2in time-slot t, 0.005
as correction factor of temperature, t0 ðtÞ as tempera­ here l as less caloric gain of biogas, @e as availability, γe
ture of outside in0C at tome-slot t, 25°C as standard as efficiency, qbr ðtÞ as recovery of biogas, ebio ðtÞ as
room temperature. manufacturing electricity in kWh in slot t.
It can be proven using the Weibull probability The net micro-CHP generator power pchp is com­
density function, and it is computed below: puted as,
� �� � � �α1
α1 ir ðtÞ ðα1 1Þ ir
β ebio ðtÞ
f ðIr ðtÞÞ ¼ ζ e 1 þ ð1 ζÞ pchp ðtÞ ¼ (15)
β1 β1 � � t � cf
� �� �
α2 ir ðtÞ ðα2 1Þ ir
β α2
� e 1 0 < ir ðtÞ < 1 where cf as factors of plant access and it is in the range
β2 β2
of [80%-90%], t as time in hours.
(11)
where ζ as factor of balancing, α1 and α2 as factors of
shape, β1 and β2 as scaling factor. 3.7. EV on-board PV
Energy produced by the electric vehicle’s on-board
3.5. Wind generation system photovoltaic system is used to power the vehicle and
Meteorological conditions and wind flow direction have its loads. Let Gt be amount of energy the electric
the most effects on the generation of wind power. An vehicle onboard PV system harvested, which is pro­
estimate amount of energy created by WT is: vided below:
( )
1 Xt
pwd ðtÞ ¼ � ρ � atb � v3s ðtÞ (12)
2 pt ¼ min LMτ 1τ ðΔω ; τÞ; Gt (16)
τ¼0
here ρ as density of air, atb as area of turbine blade,
vs ðtÞ as air speed on m/s. where 1τ ðΔω ; τÞ as whether the load LMτ is delivered
Pt
at the scheduled time, LMτ 1τ ðΔω ; τÞ as loads of
3.6. Micro-chip τ¼0
electric vehicle received at time slot t. If there con­
To satisfy its energy needs, the building also uses tains any energy pt , it must be kept in the electric
electricity from their site-built controlled micro-CHP vehicle storage, and it is used at the peak price
system. Average fuel consumption for CHP system periods while the battery energy of home is
power generation is 32% lower and reduces carbon inadequate.
emissions by 50%. Examples of renewable fuels that
can further reduce carbon emissions include renew­
able hydrogen (RH), biogas, as well as renewable nat­
3.8. Power consumption
ural gas (RNG) (Kim et al. 2015). The following
equation is used to calculate the effectiveness of There are static appliances in a smart house and smart
a CHP system: schedulable appliances, which use the energies, like
P renewable energy sources, EG, and battery storage
ωe þ qth
@e ¼ (13) system.
qf
where @e as effectiveness of net system, ωe as efficient
electric output, qth as net thermal output, and qf as 3.9. Load consumption model
input of fuel energy.
Consider there are 2 basic appliance sets of smart-
An efficient CHP system ranges from 60 to 80 per­
home, i.e, lsch and lnsch . From which, lsch refers set of
cent. For CHP, biomass is mainly made from trees,
schedulable home-uses at off peak hours, and
and it is a key fuel source. CH4 makes up a large
lnsch ðtÞ ¼ fq1 ; q2 ; q3 ; . . . ::; qn g over scheduled period
portion of biogas among 30 to 65 % of the gas is
length of t ¼ f1; 2; 3; . . . :; 24g.
CH4. The primary component of biogas is CH4, and
it can be used as a fuel for transportation to produce !
X
24 X
M �
heat and electrical energy. Equation (14) explains how sch
l ðtÞ ¼ sch
l t; m 2 M ¼ lsch t1 ; m 2 M
power can be produced from biogas. This formula t¼1 m¼1
relates the biogas transformation into useable energy þlsch t2 ; m 2 M þ . . . : þ lsch t24 ; m 2 Mg (17)
on heat and electricity form.
6 C. P. SHIRLEY ET AL.

! �
X
24 X
M � 1 if present at house
l nsch
ðtÞ ¼ sch
l t; n 2 N ¼ lnsch t1 ; n 2 N EVðAÞ ¼ (24)
0 if not present
t¼1 m¼1
þlnsch t2 ; n 2 N þ . . . : þ lnsch t24 ; n 2 Ng 8
< 1 if battery ch arg e
(18) EVðCÞ ¼ 0 if idle (25)
:
1 if disch arg e

3.10. HVAC load consumption model

3.12. Co2 emission
A building’s HVAC system provides heating, cool­
ing, and ventilation. HVAC ventilation operations It is calculated as,
may be postponed or moved to different times
co2 ¼ AVGðeprice ðtÞÞ=ψ � γe � M (26)
(ALHELOU). To calculate how much an HVAC
system affects the ambient temperature is given where AVGðeprice ðtÞÞ as average cost of electricity per
below: month, ψ as 0.20dollars per kWh price of electricity, γe
υe as 1.37 electricity emission factors.
ttþ1 ¼ εtt þ ð1 εÞðttout þ et Þ; for all t (19)
ac
here tt as ambient temperature of indoor, ttout as out­ 3.13. Par
side temperature, υe effectiveness of thermal conver­
sion, ac as general thermal conductivity on kW/0F, PAR is the ratio between the average net load used
over the course of planned hours and the maximum
ε ¼ e τ=ω from which ω as time constant of system.
load used during a given time interval. PAR is defined
from t = 1 to T = 24 h.

3.11. Models of storage MAXðloadtotal ðtÞÞ

PAR ¼ (27)
1
P
T
3.11.1. battery storage T loadtotal ðtÞ
At peak-hours, the residual renewable energy is kept at t¼1

reserve in battery storage system. By abiding by the

charging-depth and over-charging constraints, it
stores the energy. The below eqn describes the energy, 3.14. User comfort (UC) model
which is stored on battery at time slot t,
It is determined based on air quality, room tem­
Kd PEdch ðtÞ perature, visual comfort, and delay. The mathema­
BSSðtÞ ¼ BSSðt 1Þ þ Kd δBSS PEch ðtÞ tical formula and other details are described as
δBS
(20) follow.

here BSS as energy storage on Ah at time intervalt,Kd

3.14.1. Delay comfort
as time duration in hours, δBS as efficiency of battery,
Every home appliance’s schedule and serving time is
Kd as time duration in hours, PEch and PEdch as elec­ delay comfort and it is described below,
tricity power kW supplied to BSS from renewable P
energy sources, and battery-powered electricity in comfort jappunsch ðtÞ appsch ðtÞj
delay ¼ P (28)
kW supplying the load during time slot t. The below appsch
TOT
equation shows battery discharging and charging
where appunsch ðtÞ as time-slots assigned at unsched­
limits,
uled manner, appsch ðtÞ as assigned time-slots on
PEch ðtÞ � PEUB
ch
(21) scheduled manner, appsch TOT represents net available
scheduling-slots. The cost of energy usage and the
PEdch ðtÞ � PELB
dch
(22) delay are related to user comfort.

BSSðtÞ � BSSch
UB (23) 3.14.2. Visual comfort
It is influenced by the number of lights present as well
where PEch dch
UB PELB as upper charge and lower bound of as the time required to serve. By adjusting it, the user’s
discharging
preference is taken into account and the luminous
intensity is specified as,
3.11.2. EV storage mobile backup
nl � le ðtÞ � fs � # � m
The available electric vehicle and charging of bat­ visualcomfort ðtÞ; (29)
ar
tery as,
 AUSTRALIAN JOURNAL OF ELECTRICAL AND ELECTRONICS ENGINEERING 7

here nl as count of light appliances, visualcomfort ðtÞ as input, memory, state, addition, delay, hidden, and
interior luminance in ar lit indoor-space, le ðtÞ as the output. The input layer accepts the input, and also
energy consumption of each piece of lighting equip­ uses the output of the hidden layer. The output of
ment, fs as value of source-flux, # as usage factor, and Recurrent Enhanced RNN is determined by the con­
mas maintenance factor. jugate gradient-descent technique, and generalised
Armijo search approaches. The structure of RRNN is
3.14.3. Indoor thermal comfort depicted in Figure 2. RERNN is used in the suggested
Based on user choices, a customised HVAC system approach to anticipate load demand. The stepwise
regulates temperature amid a certain range. A single- procedure of RERNN is defined as,
individual must have a pleasant indoor-temperature
between 20 and 25 degree Celsius to live in a smart Step 1: Initialisation
home. Initiate the parameters, like wind speed and solar
tout;tþ1 tin;t
radiation, hourly load data, temperature etc.
rqeq hqt Step 2: Random Generation
tc;tþ1 ¼ tin;t þ (30)
ma � cp Input parameters are created at random by using
random vector.
where tc as indoor temperature, tin as outdoor tem­
Step 3: Verify the iteration is maximum
perature, hqt as tuning consumption of HVAC, rqeq as
If the process iteration is fewer than the maximal
room equivalent thermal, cp as air heat capacity, and iteration, the data process is terminated.
ma as indoor air quality. Step 4: Find Learning Rate by General Armijo Search
Method
3.14.4. Air Quality Comfort (AQC) The learning rate is defined based on generalised
The environment’s carbon-emissions are measured to Armijo search, and the equation is stated as,
gauge AQC. An adaptive ventilation system may be Step 5: Determination of Learning Rate
used to adjust the indoor carbon content in the smart The general Armijo search is used for learning rate
home. In order to preserve the indoor space may be determination and is given below,
ventilated with fresh air depending on user � �
preferences. e N K þ SR RK � e N K þ K1 SR eKw ðRK ÞT K1 � 0
(32)
ζ t þ fair ðζ out ζ t Þ þ ζ in
ζ tþ1 ¼ (31)
V Step 6: New Weight Estimation
where carbon-emission concentration as ζ tþ1 in an The gradient descent method is used to accomplish
enclosed region of volume v, which is adjusted by the new weight, and it is described as,
modifying the fresh air fair in area to the accumulated N Kþ1 ¼ N K þ Sr RK (33)
value of carbon concentration, ζ t as CO2 concentra­
tion of outdoor, ζ out , ζ in as indoor CE. Step 7: Check Maximal Iteration
If iteration is extended, then stop the method.
Otherwise, incremen the iteration, and move to the
4. Proposed CS-RERNN based intelligent step 6.
controller fed electric drive Step 8: Determine Direction
It is considered by the conjugate gradient descent
The surrounding systems are growing at the same time
method, and it is defined as
when the human life becomes more complex, resulting
in increased complexity and cost. The CS algorithm is RK ¼ eKu þ #RK 1
(34)
a geometry-based metaheuristic algorithm. RERNN pos­
sesses selective memory. CS algorithm based on energy KeK ðRK 1 Þ
T

management. An enhanced conjugate method using # ¼ K 1w K 1 ; K 2 ð0; 1Þ (35)

R ðR eKu Þ
generalised Armijo search strategy accelerates conver­
gence rate is employed to train the RERNN model. Then, move to step 4 and repeat the method again
until get better output.

4.1. RERNN method

4.2. Circle search algorithm (CS)
RERNN is a training approach that uses radial-based
functions. It is controlled via synthetic neural network. A brand-new metaheuristic optimisation technique
The difference between the Elman RNN and RERNN called the CSA and is inspiried by the geometrical
are refers to the number of levels. RERNN has six propoerties of circles (Tao et al. 2020). The circle,
layers, while the Elman recurrent neural network has which has a diameter, a perimeter, a centre, and tan­
3 layers (Qais et al. 2022). The RENNN layers include gent lines, is the most weel-known gometric shape.
8 C. P. SHIRLEY ET AL.

Figure 2. Structure of RERNN approach.

The behaviour of the CS algorithm is based on the The search agent position Yt is updated based on the
exploitation and exploration and is significantly influ­ calculated best position YC portrays on Equation 36:
enced by this features. As a result, metaheuristic algo­
rithms have emerged as a feasible solution to difficult Yt ¼ YC þ ðYC Yt Þ � tanðθÞ (39)
issues. The CS algorithm is used to solve a wide variety where the angle θ performs critical at exploration and
of optimisation issues. exploitation of CS and it may be computed as below:
�
Step 1: Initialisation ω � Rand iter > ðC � MaxiterÞ
θ¼ (40)
This phase is crucial in CS algorithm because, as ω�p otherwise
shown in Algorithm 1, each search agent’s entire
dimensions should be evenly randomised. It divides
ω ¼ ω � Rand ω (41)
the dimensions that irregularly results in surprisingly,
the algorithms produce the best outcomes. According � �2
to Equation 36, the search agents are configured to iter
A¼π π� (42)
work between the upper and lower limits of the search max iter
space. The output of demand response generated by
RERNN, power generated by the sources, cost func­ � �0:5
tion, and the related generation limit. iter
p ¼ 1 ¼ 0:8 � (43)
max iter
Yt ¼ LV þ R � ðUV LVÞ (36)
here max iter denotes maximal number of iterations,
where R as random vector that lies among [0, 1]. Rand represents random number lies amid [0,1].
Step 2: Random Solution Equation (40) portrays variable ω changes π to 0
After initialisation, the input parameters are generated through the maximisation of iterations number.
at random. Equation (41) shows the variable A changes from πto
2 3 0. The variable p changes 1 to 0 as in equation (42).
Z11 Z12 . . . : Z1d
6 Z21 Z22 . . . :: Zd 7 Step 5: Exploration and Exploitation:
I¼6 4...:
7 (37) Exploration: iter > ðC: max iterÞ means makes angle
...:5
N1 N2 . . . :: Nd θ ¼ ω � Rand all the time that may use to enhance
the exploration method of the CS.
Step 3: Fitness Function Exploitation: iter < ðC: max iter Þ means makes angle
The fitness is evaluated based on the objective func­ θ ¼ ω � p all the time may use to enhance the exploi­
tion, and it is defined as, tation method of the CS.
Step 6: Termination Criterion:
F ¼ MIN ðCost; PARÞ (38)
Verify the termination criterion, if it satisfies the con­
Step 4: Updating Search Agent Position dition means optimum solution is attained, else, recap
 AUSTRALIAN JOURNAL OF ELECTRICAL AND ELECTRONICS ENGINEERING 9

the procedure. Flowchart of CS algorithm is portrays increased to 0.5C at 5 hour. The continuous cycle of
on Figure 3. period it is end at the temperature of 1.3 °C at 24 hour.
Analysis of Biogas QbrEV availability portrays on
Figure 5. Figure 5(a) portrays analysis of Biogas Qbr
5. Results and discussion
availability. The Qbr is starting at 8(m3/h) at 1 hour.
An efficient energy management on SG considering After the many continuous cycle the Qbr is end at 4
MG with day-ahead energy forecasting, in this paper (m3/h) at 24 hour. Analysis of EV availability is shown
utilised the CS-RERNN approach. The objectives of in Figure 5(b). The EV is staring at 1 EV at 9 hour.And
the proposed work are used to decrease the cost of the EV is in 1 EV at 14 hour. The continuous cycle,
electricity, and PAR and increasing consumer com­ this EV is end at 1 EV in 24 hour. Analysis of Solar an
fort. The proposed method is analysed in different CHP power generation portrays on Figure 6. Analysis
cases such as utilising energy from EG and utilising of solar power generation portrays on Figure 6(a).
energy from EG and RES. The proposed hybrid About 90% of the solar generation is staring at 0Wh
approach will be actualised in MATLAB platform, in 1 hour, and it is increased to 70Wh at 7 hour. Solar
and evaluated their performance. generation is decreased at 82Wh at 17 hour and the
Analysis of Forecasted temperature shown in solar generation is end at 0Wh at 24 hour. Analysis of
Figure 4. Figure 4(a) shows the analysis of ambient CHP power generation is shown in Figure 6(b).
temperature. The temperature is starting in 17°C at Estimated u-chip generation, the power is starting in
1 hour. After the continuous cycle the temperature is 102Wh at 1 hour, and it is increased in 118 Wh at
ended at 18 °C at 24 hour. Analysis of forecasted wind 3 hour, then it is increased at 95 Wh in 5 hour, and
flow speed portrays on Figure 4(b). The temperature is again it is inceased at 105 Wh at 13 hour. After the
starting in 0.4 °C at 2 hour and the temperature is continous cycle, the power is end at 42 Wh at 24 hour.

Figure 3. Flowchart of CS algorithm.

10 C. P. SHIRLEY ET AL.

Analysis of Wind power generation portrays on 0.28, CSA is at 0.28, BCO is at 0.4. In TV proposed is
Figure 7. In Estimated WE, the power is starting at 1 at 0.1, PSO is at 0.1, CSA is at 0.44, BCO is at 0.52. In
Wh at 1 hour, and it is increased at Wh 1 at 5, and it is DW proposed is at 0.06, PSO is at 0.08, CSA is at 0.42,
increased at Wh 22 at 7 hour. In remaining WE after BCO is at 0.6. In MO proposed is at 0.02, PSO is at
BSS charging, the power is starting at 1 Wh at 1 and 0.02, CSA is at 0.3, BCO is at 0.62. From this compar­
the wind energy is increased power at 18 Wh in 11, ison of waiting time of the entire appliances in FP in
and after some oscillation of time, the power is end at proposed takes low time to do work. Comparison of
15 Wh in 24 hour. Analysis of battery storage level EV waiting time of the entire appliances at DAP is shown
battery charging level is depicts in Figure 8. Analyses in Table 2. In WH proposed is at 0.08, PSO is at 0.38,
of battery storage level is depicts on Figure 8(a). In CSA is at 0.38, BCO is at 0.5. In Refr proposed is at
Battery charging level is started at 0 Ah at 2 hour and 0.48, PSO is at 0.5, CSA is at 1. BCO is at 1. In HVAC
they increased at 80 Ah at 7 hour. After the continuous proposed is at 0.08, PSO is at 0.1, CSA is at 0.1, BCO
cycle the batter storage level is end at 280Ah at 24. is at 1.01. In WM proposed is at 0.04, PSO is at 0.5,
Analysis of EV battery charging level is shown in CSA is at 0.24, BCO is at 0.48. In EV proposed is at
Figure 8(b). It is starting from 0 Ah at 200 hour and 0.08, PSO is at 0.3, CSA is at 0.03, BSO is at 0.32. In
after some period at 80 Ah at 15 hour and after some Lights proposed is at 1, PSO is at 1, CSA is at 0.32,
period it is end at 50 Ah at 24 hour. BCO is at 0.58. In TV proposed is at 0, PSO is at 0.32,
Comparison of waiting time of the entire appli­ CSA is at 0.32, BCO is at 0.52. In DW proposed is at
ances is shown in Table 1. In WH proposed is at 0.38, 0.5, PSO is at 0.52, CSA is at 1, BCO is at 1.4. In MO
PSO is at 0.38, CSA is at 0.38, BCO is at 0.5. In Refr proposed is at 0.08, PSO is at 0.9, CSA is at 0.44, BCO
proposed is at 0.4, PSO is at 0.56, CSA is at 1, BCO is is at 0.56. Comparison of waiting time of all appli­
at 1. In HVAC proposed is at 0, PSO is at 0.36, CSA is ances in DAP in proposed takes low time to do work.
at 0.36, BCO is at 0.62. In WM proposed is at 0, PSO Comparison of thermal comfort and quality comfort
is at 0.01, CSA is at 0.01, BCO is at 0.01. In EV is shown in Table 3. In Proposed the temperature is
proposed is at 0.48, PSO is at 0.5, CSA is at 0.88, starting from 24.2 °C at 3 hour. Again, the tempera­
BCO is at 1. In Lights proposed is at 0.2, PSO is at ture is at 25 °C at 8 hour, then the cycle is continuous

Figure 4. Analysis of forecasted (a) ambient temperature (b) wind flow speed.

Figure 5. Analysis of (a) biogas Qbr availability (b) analysis of EV availability at home.
 AUSTRALIAN JOURNAL OF ELECTRICAL AND ELECTRONICS ENGINEERING 11

Figure 6. Analysis of (a) solar power genertion(b) CHP power generation.

4 hour at the next cycle the line is decreased at 745 °C

in 6 hour and after a continous cycle the line end at
735 °C at 24 hour. In PSO, the line is start from 762
°C at 1 hour then it is decreased at 745 °C at 3 hour
and it is increased at 762 °C at 5 hour after the
continous cycle the line is end at 158 °C at 24 hour.
In CSA the line is starting at 162 °C at 1 hour, then it
is decreased at 745 °C at 2 hour, and it is again
increased at 162 °C at 5 hour after a continous cycle
the line is end at 159 °C at 24 hour. In BCO the line
started fron 152 °C at 1 hour it is slightly decreased at
Figure 7. Analysis of wind power generation. 149 °C at 3 hour at the temperture 145 °C at 8 after
the continuous cycle the line is end in 151 °C at
24 hour. In Lower thermal bond, the value is con­
to fo forward when the temperature is at 23.6 °C at stant. Comparison of Visual comfort in DAP and FP
23 hour after some continuos cycle this is end at 23.8 proposed and existing methods portrays on Table 4.
°C at 24 hour. In Proposed the temperture is starting In proposed is the line start from 330 (cd/m2) at
at 754 °C at 1 hour then it is increased at 758 °C at 1 hour and the line is decreased at 250 (cd/m2) at

Figure 8. Analyses of (a) battery storage level and (b) EV battery charging level.

Table 1. Comparison of waiting time of the entire appliances at FP with existing methods.
Solution Techniques WH Refr HVAC WM EV Lights TV DW MO
PSO 0.38 0.56 0.36 0.01 0.5 0.28 0.1 0.08 0.02
CSA 0.38 1 0.36 0.01 0.88 0.28 0.44 0.42 0.3
BCO 0.5 1 0.62 0.01 1 0.4 0.52 0.6 0.62
Proposed 0.38 0.4 0 0 0.48 0.2 0.1 0.06 0.02
12 C. P. SHIRLEY ET AL.

Table 2. Comparison of waiting time of the entire appliances at DAP with existing methods.
Solution Techniques WH Refr HVAC WM EV Lights TV DW MO
PSO 0.38 0.5 0.1 0.5 0.3 1 0.32 0.52 0.9
CSA 0.38 1 0.1 0.24 0.03 0.32 0.32 1 0.44
BCO 0.5 1 1.01 0.48 0.32 0.58 0.52 1.4 0.56
Proposed 0.08 0.48 0.08 0.04 0.08 1 0 0.5 0.08

2 hour and it again incresed at 330 at 4 hour aftre the cents at 24 hour. The proposed technique is lower cost
continous cycle the line is end at 326 (cd/m2) at comparing the existing technique. Comparisons of PAR
24 hour. In PSO the line is starting at 250 (cd/m2) in DAP and PAR in FP proposed and existing techni­
at 1 hour and it is travelled stright at 250 (cd/m2) at ques is shown in Table 6. In Unscheduled is higher in
13 hour after that that is increased at 282 (cd/m2) at proposed system, all others are low. In Scheduled sys­
13 hour and at the continuous cycle the line is end at tem proposed system is lower compared to other sys­
286 (cd/m2) at 24 hour. In Outer temperature profile tem. The proposed value is lower scheduled comparing
the luminance is started from 250 (cd/m2) at 7 hour the existing technique. Compared the scheduled line the
the it is increased to 318 (cd/m2) at 8 and then It is proposed in same level compared to others. Compared
increased at 380 (cd/m2) at 11 hour then costant line the scheduled line the proposed is lower compared to
liminance at 380 (cd/m2) at 16 hour then the cycle is others. The proposed value is lower scheduled compar­
end at 250 (cd/m2) at 13 hour.In Upper thermal ing the existing technique. Comparison of Carbon
bound value is constant.In proposed the line is Emission in FP and DAP of proposed and existing
started at 250 (cd/m2) at 1 hour and then the line is techniques portrays on Table 7. In proposed Carbon
increased at 330 (cd/m2) and 2 hour and then the line Emission is started at 122 at 1 hour and after
is degreased at 250 (cd/m2) at 5 after the continous a continuous cycle the cycle is end at 122 at 24 hour.
cycle the line is end at 250 (cd/m2) at 24 hour. The carbon emission is started at 139 at 1 hour and the
line is increased to 78 at 6 hour and the line travel at
Case 1: Analysis of Proposed Method based on straight and still carbon Emission at 78 at 11 hour and
Utilising Energy from EG the continuous cycle the line is end at 140 at 24 hour.

Comparison of (a) day-ahead price and (b) Flat price in Case 2: Analysis of Proposed Method based on
proposed and existing methods is portrays on Table 5. Utilising Energy Form EG and RES
In proposed, the cost is starting at 28 cents at one hour
and the cost is reduced to 18 cents at 4 hour and the cost Comparison of electricity bill Cost portrays on
is increase at 25 cents at 5 hour and the continuous cycle Table 8. The proposed is starting from cents at
the cost is end at 22 cents at 24 hour. The proposed 70 cents at1 hours and it is end at 64 cents at
technique is lower cost comparing the existing techni­ 24 hour. In Proposed power is starting at 30 at
que. Proposed, it is starting at a cost of 60 cents at 1 hour and it is in 32 at 2 hour and it is at 24 in
1 hour and again it is degreased to 40 cents at 3 hour 10 hours after a continuous cycle the line is end at
and increased to 60 cents at 5 hour and it is end at 60 25 at 24 hour.

Table 3. Comparison of thermal comfort and quality comfort.

Thermal Comfort Quality Comfort
Solution Techniques Temperature (oC) Time (h) Temperature (oC) Time (h)
PSO 22 °C 24 158 °C 24
CSA 23 °C 24 159 °C 24
BCO 23 °C 24 151 °C 24
Proposed 23.8 °C 24 735 °C 24
Outdoor temperature profile 18 °C 24 748 °C 24
Upper thermal bound 26 °C 24 760 °C 24
Lower thermal bound 21 °C 24 739 °C 24

Table 4. Comparison of visual comfort in DAP and FP with existing methods.

DAP FP
Solution Techniques Luminance (cd/m2) Time (h) Luminance (cd/m2) Time (h)
PSO 286 (cd/m2) 24 296 (cd/m2) 24
CSA 310 (cd/m2) 24 310 (cd/m2) 24
BCO 302 (cd/m2) 24 302 (cd/m2) 24
Proposed 326 (cd/m2) 24 250 (cd/m2) 24
Outdoor temperature profile 250 (cd/m2) 17 250 (cd/m2) 13
Upper thermal bound 320 (cd/m2) 24 320 (cd/m2) 24
Lower thermal bound 290 (cd/m2) 24 290 (cd/m2) 24
 AUSTRALIAN JOURNAL OF ELECTRICAL AND ELECTRONICS ENGINEERING 13

Table 5. Comparison of (a) day-ahead price and (b) flat price in proposed and existing methods.
DAP FP
Solution techniques Electricity Bill Cost (Cents) Time (h) Electricity Bill Cost (Cents) Time (h)
PSO 25 24 62 24
CSA 24 24 70 24
BCO 28 24 60 24
Proposed 22 24 60 24

Comparison of PAR in FP and PAR in DAP in Comparison of efficiency with proposed and
proposed and existing techniques portrays on existing method is depicts in Table 11. In PSO
Table 9. Compared the UN scheduled of proposed is method, the efficiency is 68%. In CSA method,
equal compared to others. The scheduled proposed is the efficiency is 78%. In BCO method, efficiency
lower compared to other existing methods. Compared is 86%. In proposed method, efficiency is 96%.
the UN schedule of proposed is equal compared to Here, the proposed method shows a high effi­
others. The scheduled proposed is less compared to ciency compared with other existing methods.
other existing methods. Hence, the proposed method
takes lower time compared to existing techniques. Case 3: Utilising energy from EG, BSS and RESs
Comparison of Carbon Emission in DAP and FP of
proposed and existing techniques portrays on The smart house in this case uses external grid
Table 10. In proposed of carbon emission is at 101 at electricity as well as RESs, BSS, and EV storage for
1 hour, and it is degreased to 59 at 1 hour after the unscheduled and scheduled loads. In response to
continuous cycle the carbon emission is end at 118 at FP, we will analyse electricity cost, CE, PAR, and
24 hour. In Proposed of carbon Emission is started user comfort. Table 12 shows the Comparison of
at140 at 1 hour and it is slightly decrease at 60 at (a) day-ahead price and (b) Flat price in proposed
4 hour after a continuous cycle the line is end at 140 and existing methods. Table 13 displays
at 24 hour. Comparison of (a) PAR in DAPs and (b) PAR in

Table 6. Comparison of (a) PAR in DAPs and (b) PAR in FP proposed and existing methods.
PAR in DAP PAR in FP
Solution Techniques Unshaded schedule Shaded schedule Unshaded schedule Shaded schedule
PSO 2.8 2.3 2.8 2.2
CSA 2.8 2.4 2.8 2.3
BCO 2.8 2.8 2.8 2.8
Proposed 2.8 2 2.8 2

Table 7. Comparison of carbon emission in FP and DAP of proposed and existing techniques.
FP DAP
Solution Techniques Carbon Emission Time (h) Carbon Emission Time (h)
PSO 124 24 150 24
CSA 122 24 139 24
BCO 202 24 239 24
Proposed 122 24 140 24

Table 8. Comparison of electricity bill cost of FP and DAP with existing techniques.
DAP FP
Solution techniques Electricity Bill Cost (Cents) Time (h) Electricity Bill Cost (Cents) Time (h)
PSO 60 24 25 24
CSA 55 24 25 24
BCO 70 24 18 24
Proposed 64 24 25 24

Table 9. Comparison of PAR in FP and PAR in DAP with proposed and existing techniques.
PAR in FP PAR in DAP
Solution Techniques Unshaded schedule Shaded schedule Unshaded schedule Shaded schedule
PSO 2.3 1.4 2.2 1.3
CSA 2.3 1.5 2.2 1.3
BCO 2.3 1.6 2.2 1.4
Proposed 2.3 1.3 2.2 0.9
14 C. P. SHIRLEY ET AL.

Table 10. Comparison of carbon emission in DAP and FP with existing techniques.
DAP FP
Solution Techniques Carbon Emission Time (h) Carbon Emission Time (h)
PSO 139 24 250 24
CSA 122 24 190 24
BCO 200 24 260 24
Proposed 118 24 140 24

Table 11. Comparison of efficiency with proposed and existing

and evaluated their performance. Also, the smart home
methods. is connected to a grid-tied MG powered by RESs. A CS-
Solution Techniques Efficiency (%) RERNN model based on historical data is also con­
PSO 68 structed. The proposed approach is analysed under dif­
CSA 78
BCO 86
ferent cases, such as utilising energy from EG and
Proposed 96 utilising energy from EG and RES. The performance is
compared with various existing techniques. The pro­
posed method shows low electricity bill cost and lowest
FP proposed and existing methods. Table 14 energy cost, low carbon emission as compared with other
shows the Comparison of Carbon Emission in (a) existing methods. Limitations in data quality, changes in
FP and (b) DAP of proposed and existing weather patterns, and unforeseen events can lead to
techniques. inaccuracies in the predictions, which can subsequently
impact the efficiency of energy management in micro­
grids. Accurate forecasting requires high-quality data
6. Conclusion from various sources, including weather data, historical
This section concludes a CS-RERNN method for home energy consumption, and grid conditions. In the future,
energy management to reduce the energy cost of resi­ we will use real-time and multi-objective optimisation
dential household. The main aim of the proposed tech­ approaches to control the energy of several smart houses,
nique is to decrease the electricity cost and PAR and buildings, and microgrids. We will also develop systems
increasing consumer comfort. The proposed CS- for coordinating energy management between
RERNN approach is actualised in MATLAB platform microgrids.

Table 12. Comparison of (a) day-ahead price and (b) flat price in proposed and existing methods.
DAP FP
Solution techniques Electricity Bill Cost (Cents) Time (h) Electricity Bill Cost (Cents) Time (h)
PSO 20 24 76 24
CSA 22 24 60 24
BCO 11 24 53 24
Proposed 17 24 42 24

Table 13. Comparison of (a) PAR in DAPs and (b) PAR in FP proposed and existing methods.
PAR in DAP PAR in FP
Solution Techniques Unshaded schedule Shaded schedule Unshaded schedule Shaded schedule
PSO 1.9 1.6 1.9 1.8
CSA 1.9 1.75 1.9 1.7
BCO 1.9 1.3 1.9 1.21
Proposed 1.9 1.2 1.9 1.25

Table 14. Comparison of carbon emission in (a) FP and (b) DAP of proposed and existing techniques.
FP DAP
Solution Techniques Carbon Emission Time (h) Carbon Emission Time (h)
PSO 200 24 155 24
CSA 160 24 90 24
BCO 140 24 100 24
Proposed 130 24 110 24
 AUSTRALIAN JOURNAL OF ELECTRICAL AND ELECTRONICS ENGINEERING 15

Disclosure statement Scheduling with Reinforcement Learning.” IEEE

Transactions on Smart Grid 7 (5): 2187–2198.
No potential conflict of interest was reported by the author(s). Moon, S., and J. W. Lee. 2016. “Multi-Residential Demand
Response Scheduling with Multi-Class Appliances in
Smart Grid.” IEEE Transactions on Smart Grid 9 (4):
References 2518–2528.
Qais, M. H., H. M. Hasanien, K. H. L. Saad Alghuwainem,
Alhelou, H. “An Optimal Power Usage Scheduling in Smart M. A. Elgendy, and R. A. Turky. 2022. “Accurate
Grid Integrated with Renewable Energy Sources for Three-Diode model estimation of Photovoltaic modules
Energy Management.” using a novel circle search algorithm.” Ain Shams
Arun Kumar, A., B. V. Manikandan, S. Kannan, and Engineering Journal 13 (3): 101824.
A. Bhuvanesh. 2023. “Recuperation Based Rajani, B., and B. Naidu Kommula. 2022. “An Optimal
Multi-Objective Generation Expansion Planning for a Energy Management Among the Electric Vehicle
Real-World Power System.” Journal of Intelligent & Charging Stations and Electricity Distribution System
Fuzzy Systems Preprint (4): 1–16. https://doi.org/10. Using GPC-RERNN Approach.” Energy 245:123180.
3233/JIFS-232909. https://doi.org/10.1016/j.energy.2022.123180.
Aslam, S., Z. Iqbal, N. Javaid, Z. Ali Khan, K. Aurangzeb, Rehman, A. U., G. Hafeez, F. R. Albogamy, Z. Wadud,
and S. Irtaza Haider. 2017. “Towards Efficient Energy F. Ali, I. Khan, G. Rukh, and S. Khan. 2021. “An
Management of Smart Buildings Exploiting Heuristic Efficient Energy Management in Smart Grid
Optimization with Real Time and Critical Peak Pricing Considering Demand Response Program and
Schemes.” Energies 10 (12): 2065. Renewable Energy Sources.” Institute of Electrical and
Aslam, S., A. Khalid, and N. Javaid. 2020. “Towards Efficient Electronics Engineers Access 9:148821–148844. https://
Energy Management in Smart Grids Considering doi.org/10.1109/ACCESS.2021.3124557.
Microgrids with Day-Ahead Energy Forecasting.” Shuja, S. M., N. Javaid, M. Zeeshan Rafique, U. Qasim,
Electric Power Systems Research 182:106232. https://doi. R. Farhat Makhdoom Khan, A. Anjum Butt, and
org/10.1016/j.epsr.2020.106232. M. Hanif. 2020. “Towards Efficient Scheduling of Smart
Esther, B. Priya, and K. Sathish Kumar. 2016. “A Survey on Appliances for Energy Management by Candidate
Residential Demand Side Management Architecture, Solution Updation Algorithm in Smart Grid.” In
Approaches, Optimization Models and Methods.” Advanced Information Networking and Applications:
Renewable and Sustainable Energy Reviews 59:342–351. Proceedings of the 33rd International Conference on
https://doi.org/10.1016/j.rser.2015.12.282. Advanced Information Networking and Applications
Farinis, George Κ., and Fotios D. Kanellos. 2021. “Integrated (AINA-2019) 33, pp. 67–81. Springer International
energy management system for Microgrids of building Publishing.
prosumers.” Electric Power Systems Research 198:107357. Taherian, H., M. Reza Aghaebrahimi, L. Baringo, and
https://doi.org/10.1016/j.epsr.2021.107357. S. Reza Goldani. 2021. “Optimal Dynamic Pricing for an
Garcia-Torres, F., C. Bordons, J. Tobajas, J. José Márquez, Electricity Retailer in the Price-Responsive Environment
J. Garrido-Zafra, and A. Moreno-Muñoz. 2020. “Optimal of Smart Grid.” International Journal of Electrical Power
Schedule for Networked Microgrids Under Deregulated & Energy Systems 130:107004. https://doi.org/10.1016/j.
Power Market Environment Using Model Predictive ijepes.2021.107004.
Control.” IEEE Transactions on Smart Grid 12 (1): Tao, G., X. Gong, K. Zhang, F. Lin, J. Wang, T. Huang, and J.
182–191. M. Zurada. 2020. “A Recalling-Enhanced Recurrent
Gnanaprakasam, C. N., S. Meena, M. Nivethitha Devi, Neural Network: Conjugate Gradient Learning
N. Shanmugasundaram, and S. Sridharan. 2023. “Robust Algorithm and Its Convergence Analysis.” Information
Energy Management Technique for Plug-In Hybrid Sciences 519:273–288. https://doi.org/10.1016/j.ins.2020.
Electric Vehicle with Traffic Condition Identification.” 01.045.
Applied Soft Computing 133:109937. https://doi.org/10. Ullah, K., G. Hafeez, I. Khan, S. Jan, and N. Javaid. 2021. “A
1016/j.asoc.2022.109937. Multi-Objective Energy Optimization in Smart Grid with
Gungor, V. C., D. Sahin, T. Kocak, S. Ergut, C. Buccella, High Penetration of Renewable Energy Sources.” Applied
C. Cecati, and G. P. Hancke. 2011. “Smart Grid Energy 299:117104. https://doi.org/10.1016/j.apenergy.
Technologies: Communication Technologies and 2021.117104.
Standards.” IEEE Transactions on Industrial Informatics Vosoogh, M., M. Rashidinejad, A. Abdollahi, and
7 (4): 529–539. M. Ghaseminezhad. 2020. “An Intelligent Day Ahead
Gungor, V. C., D. Sahin, T. Kocak, S. Ergut, C. Buccella, Energy Management Framework for Networked
C. Cecati, and G. P. Hancke. 2012. “Smart Grid and Smart Microgrids Considering High Penetration of Electric
Homes: Key Players and Pilot Projects.” IEEE Industrial Vehicles.” IEEE Transactions on Industrial Informatics
Electronics Magazine 6 (4): 18–34. 17 (1): 667–677.
Huaman, R. N. E., and T. Xiu Jun. 2014. “Energy Related Zhao, J., Z. H. Guo, S. Zhong-Yue, Z. Y. Zhao, X. Xiao,
CO2 Emissions and the Progress on CCS Projects: A and F. Liu. 2016. “An Improved Multi-Step
Review.” Renewable and Sustainable Energy Reviews Forecasting Model Based on WRF Ensembles and
31:368–385. https://doi.org/10.1016/j.rser.2013.12.002. Creative Fuzzy Systems for Wind Speed.” Applied
Kim, B. G., Y. Zhang, M. Van Der Schaar, and J. W. Lee. Energy 162:808–826. https://doi.org/10.1016/j.ape
2015. “Dynamic Pricing and Energy Consumption nergy.2015.10.145.