Energy Reports 7 (2021) 314–323

Contents lists available at ScienceDirect

Energy Reports
journal homepage: www.elsevier.com/locate/egyr

A grasshopper optimization algorithm for optimal short-term

hydrothermal scheduling
∗
Xie Zeng a,b ,1 , Ali Thaeer Hammid c ,1 , Nallapaneni Manoj Kumar d , ,
Umashankar Subramaniam e , Dhafer J. Almakhles e
a
School of Computer Science, Baoji University of Arts and Sciences, Baoji, 721000, China
b
Department of Computing, Faculty of Arts, Computing and Creative Industry, Universiti Pendidikan Sultan Idris, Malaysia
c
Computer Engineering Techniques Dep., Faculty of Information Technology, Imam Ja’afar Al-Sadiq University, Baghdad, Iraq
d
School of Energy and Environment, City University of Hong Kong, Kowloon, Hong Kong
e
Department of Communications and Networks, Renewable Energy Lab, College of Engineering, Prince Sultan University, Riyadh 12435, Saudi Arabia

article info a b s t r a c t

Article history: The optimal generation for short-term hydrothermal scheduling (OGStHS) with the deliberation of
Received 31 March 2020 various purposes is a complex non-linear constrained optimization problem. There exist numerous
Received in revised form 26 October 2020 constraints, which make the OGStHS optimization problem more complicated. The considered con-
Accepted 23 December 2020
straints for this problem are mostly related to energy performance, operational conditions, water,
Available online 4 January 2021
and power infrastructure. All these constraints would generally influence the cost of fuel. In this
Keywords: study, a multi-objective optimization form of OGStHS is suggested to estimate the minimum cost
Thermal power plants of fuel, which mainly influences industrial operation. The water transfer delays among multi-related
Co-ordinated hydrothermal reservoirs and the thermal plants’ valve-point influences are considered for the accurate formulation of
Hydrothermal constraint the OGStHS problem. Meantime, a grasshopper optimization algorithm (GOA) is performed to handle
Water transfer delays the OGStHS problem by getting optimized for both objectives concurrently. A modern approach is
Valve-point influences shown in this study to get a solution to the OGStHS problem. Furthermore, to deal with the complex
Grasshopper optimization algorithm
restraints efficiently, modern heuristic restriction treatment processes with no drawback impact frames
have been offered in this study. Two hydrothermal power systems have illustrated the suggested
GOA technique’s utility and performance. Compared with other available approaches, the analytical
results are admitted that GOA can provide a better understanding by decreasing fuel cost and emission
concurrently.
© 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction not been as significant as thermal plants (Kumar and Mohan,

2011; Lu et al., 2011; Hammid et al., 2018a). Therefore, the
The short-term hydrothermal scheduling (StHS) is the most OGStHS problem’s purpose was to minimize the entire cost of
concerning problem for ensuring an interrelated power system’s the thermal plant’s fuel input while performing complex environ-
optimal operation. In the hydrothermal power system (HPS), the mental constraints and operations limitations. These include the
StHS problem refers to the optimal generation for short-term balance of load and water dynamics and constraints of reservoir
hydrothermal scheduling (OGStHS). This indicates locating the storage volumes and water discharge limitations and production
optimal volume of water discharges of hydro plants and thermal capacity limitations. Regularly, in HPS, the fuel cost function for
plants’ power resources during a schedule extent to minimize an individual thermal unit is described nearly as a quadratic
the entire energy production cost while performing different function. The energy production for an individual hydro unit has
hydraulic constraints and operations limitations. It is evident that, been defined by stream and reservoirs storages of water con-
in an interrelated HPS, hydropower plant’s operation cost has currently. Therefore, the probable water resources available for
energy production through hydropower turbines at every period
∗ Corresponding author. of the schedule extent based on the prior scheduling decision.
E-mail addresses: [email protected] (X. Zeng), The dynamic combination generally sets these decisions among
[email protected], [email protected] (A.T. Hammid),
[email protected], [email protected] (N.M. Kumar),
the system’s operators based on the features during the schedule
[email protected] (U. Subramaniam), [email protected] extent. Analytically, the OGStHS problem may be classified as
(D.J. Almakhles). an extensive, robust, nonlinear, and non-convex optimization
1 These authors contributed equally to this work. problem with different complex constraints, determining the final

https://doi.org/10.1016/j.egyr.2020.12.038
2352-4847/© 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
X. Zeng, A.T. Hammid, N.M. Kumar et al. Energy Reports 7 (2021) 314–323

decision optimal schedule with system challenges efficiently (Lu 2. Problem formulation
et al., 2010; Hammid et al., 2018b; Hammid and Sulaiman, 2018).
Over the years, several conventional approaches have been The optimal generation for OGStHS with the deliberation of
suggested to make a solution to the OGStHS problem. The conven- various purposes is a complex non-linear, constrained optimiza-
tional approaches are mostly mathematical techniques such in- tion problem.
clude linear programming (LP) (Piekutowski et al., 1993), decom- Hence, the objective function and the associated constraints
position method (Mohan et al., 1992), progressive optimality al- are formulated, and the details are presented in the below sub-
gorithm (POA) (Turgeon, 1981), dynamic programming (DP) (Tang sections.
and Luh, 1995; Chang et al., 1990), Lagrange relaxation-based
algorithms (Guan et al., 1995; Salam et al., 1998). However, 2.1. Objective function
most of these studies do not have the capability to determine
optimal scheduling decisions due to their shortcomings. While The principal goal of the OGStHS problem is getting minimized
the suffering, drawback, and inability of the above highlighted, the gross cost of fuel (F ) while performing different types of
conventional approaches are briefly explained in ref. Lu et al. constraints. The objective function for generating units (Ns ) with
(2010). interval times (T ) is expressed by Eq. (1) (Das and Bhattacharya,
Several recent heuristics stochastic search techniques include 2018):
genetic algorithm (GA) (Orero and Irving, 1998; Yuan et al., 2002),
T Ns
evolutionary programming (Sinha et al., 2003; Basu, 2004), sim- ∑ ∑ ( t)
ulated annealing (Wong and Wong, 1994), particle swarm opti- F = min fi Psi (1)
t =1 i=1
mization (PSO) (Mandal et al., 2008; Hota et al., 2009; Zhang et al.,
2016), differential evolution (DE) algorithm (Lakshminarasim- where the energy production of the ith thermal plant is (Psit ) at
man and Subramanian, 2008; Mandal and Chakraborty, 2008), a specific period of time (t). The cost of fuel for the ith unit
artificial bee colony algorithm (Tehzeeb ul et al., 2020), Gray is ((fi (Psi )) at the production of the Psit , which is determined by
Wolf Optimization (Sharma et al., 2020), mixed-integer formu- Eq. (2):
lation of the LP (Jian et al., 2019), An adaptive cuckoo search
fi (Psi ) = xi + (yi · Psi ) + (zi · Psi2 ) + ai sin bi · Psi,min − Psi
( ( ))
algorithm (Nguyen et al., 2018), Teaching learning-based op- (2)
timization (TLBO) (Roy, 2013), and Improved harmony search Moreover, in the real world HPS, there will be an apparent rise
algorithm (Nazari-Heris et al., 2018). All these algorithms can in the thermal fuel price, which would influence the overall fuel
show their ability to find out the solution to the OGStHS problem cost curve. Besides, the wire design also affects when the valve’s
without limitation on the problem’s non-convex aspect due to steam entry begins to start. The apparent loss of fuel rise has
their robust capability for searching. Nonetheless, rather than been expressed as valve point influences. The curve of fuel cost
supplying the final decision of the globe optimal schedule, all of the thermal plant with the main idea of valve-point influences
the above-highlighted techniques can give a suboptimal solu- is quite exact to the equation described in Eq. (2). where xi , yi ,
tion for obtaining the restricted local optimal points since their and zi are the coefficients of cost for generator i. Also, ai , bi are
shortcomings. the valve-point influences coefficients of the generator i. Psi,min is
The No Free Lunch hypothesis in the optimization field the minimum power generation limit of generator i.
(Wolpert and Macready, 1997) says that there is not and will not
be an optimization technique to solve entire optimization prob- 2.2. Constraints
lems. Consequently, a recent strategy based on the grasshopper
optimization algorithm (GOA) can solve the current subproblem
The associated constraints of the OGStHS problem are briefly
on a collection of problems. The GOA, which appeared recently,
described in the below sections (Das et al., 2018; Hammid et al.,
may have the ability to determine the final decisions more re-
2017):
liably with or without justification when comparing the current
techniques. Saremi et al. (2017) and Aljarah et al. (2018) proved
2.2.1. Generation of hydropower
that the approach mentioned above exceeded the extreme of the
For the jth hydro plant at a specific period of time (t), power
existing techniques on problems include system challenges with a
generation is determined by Eq. (3):
large-scale amount of local solutions (multi-modal). This modifi-
)2 ( )2 )2
cation on the engagement of the approach in this cited study aims t
= z1j Vhjt + z2j Qhjt + z3j Vhjt Qhjt + z4j Vhjt + z5j Qhjt + z6j (3)
( (
Phj
to obtain its recent operators’ utility to make a solution and be
t
adequately handled by the OGStHS optimization problem’s con- where the power production of the jth hydro plant is (Phj ) at
straints. Lastly, from the parameters of various constraints of the a specific period of time (t). z1j , z2j , z3j , z4j , z5j , and z6j are the
OGStHS problem, the procedures governing constraints related to coefficients. Vhjt is the volume of the jth hydro plant at a specific
the heuristic approaches without utilizing any discipline features period of time (t). Qhjt is the water release of the jth hydro plant
are suggested for GOA to handle the problem’s constraints. at a specific period of time (t).
Hence, in this study, to prove the suggested approach’s prob-
ability and good performance, i.e., GOA has been employed to 2.2.2. Balance constraints of real energy
make a solution for the OGStHS problem of two different hy- The constraint related to real energy is given by Eq. (4):
drothermal systems. The article is structured in six different Ns Nh
sections; in Section 2, the problem formulation is briefly pre- ∑ ∑
Psit + t
Phj = PDt + Plossj (4)
sented. Section 3 presents the GOA use for the OGStHS problem.
i=1 j=1
In Section 4, the optimization’s execution procedure is discussed,
and results are briefly explained along with the literature com- where the gross demand for the load is (PDt ) at a specific period
parison in Section 5. Finally, the conclusions were drawn in of time (t). The number of thermal and hydro plants is (Ns ) and
Section 6. (Nh ), respectively.
315
X. Zeng, A.T. Hammid, N.M. Kumar et al. Energy Reports 7 (2021) 314–323

2.2.3. Operating limitations of thermal plant 3. Grasshopper optimization algorithm for hydrothermal
The constraint related to the operating limits of the thermal scheduling
plant is given by Eq. (5):
The GOA algorithm has been suggested to address the problem
Psi,min ≤ Psit ≤ Psi,max of OGStHS, and the more detailed information on how it is applied
i = 1, 2, . . . , Ns can be found in refs. Saremi et al. (2017) and Aljarah et al. (2018).
This algorithm simulates the normal behavior of grasshopper’s
t = 1, 2, . . . , T (5) swarms. The trajectory of jumping off for every grasshopper in
where the minimum and the maximum energy production of the a specific swarm is influenced by three factors: group interac-
ith thermal plant are Psi,min and Psi,max respectively. tion (Gi ), gravity (Gr ), and wind effect horizontally (We ). In the
GOA algorithm, group interaction is the principal procedure of
exploration determined according to Eq. (11):
2.2.4. Operating limitations of the hydro plant N
The constraint related to the operating limits of the hydro ∑ ( )
Gi = G lij l̂ij (11)
plant is given by Eq. (6):
j=1,j̸ =i
t
Phj,min ≤ Phj ≤ Phj,max where the length between ith and jth grasshopper is (lij ) then it
j = 1, 2, . . . , Nh has been determined as lij = xj − xi , G used to determine the
x −x
intensity of group effectiveness, and l̂ij = j l i is an element
t = 1, 2, . . . , T (6) ij
trajectory from ith and jth grasshopper.
where the minimum and the maximum energy production of the As mentioned above, the principal part of group interaction is
jth hydro plant are Phj,min and Phj,max respectively. the use of G. This use describes the motion orientation of each
grasshopper in the swarm and determined according to Eq. (12):
2.2.5. Limitations of water release for hydro plants
−r
The constraint related to the limitations of water release for S(r) = ke d − e−r (12)
the hydro plant is given by Eq. (7):
where the strength of attraction is (k), and the attractive distance
system is (d).
Qhj,min ≤ Qhjt ≤ Qhj,max The above function produces two kinds of powers between
each grasshopper: aversion and attraction. When the distance
j = 1, 2, . . . , Nh
between every two grasshoppers is in [0, 2.079], they repulse
t = 1, 2, . . . , T (7) each other to prevent a clash. The power of attraction grows if
the distance is between [2.079, 4] to keep the swarm’s coherence.
where the minimum and the maximum water release of the jth If the distance is precisely 2.079, there is no power, and this state
hydro plant are Qhj,min and Qhj,max respectively. is named the rest area.
The simulation in interactions within grasshoppers produces
2.2.6. Volumes limitations of reservoir storage a useful swarm model. Moreover, it should be modified to create
The constraint related to the limitations of the volume of an optimization approach. Related to refs. Saremi et al. (2017),
and Aljarah et al. (2018) suggested the subsequent arithmeti-
reservoir storage is given by Eq. (8):
cal design exploration while grasshoppers were interacting. The
Vhj,min ≤ Vhjt ≤ Vhj,max arithmetical design has been described according to Eq. (13):
j = 1, 2, . . . , Nh
⎛ ⎞
N
∑ ubl − lbl (⏐ l ⏐) xj − xi
t = 1, 2, . . . , T (8) Xil = c ⎝ c s ⏐xj − xli ⏐ ⎠ + T̂d (13)
s lij
j=1,j̸ =i
where the minimum and the maximum volumes of the jth hydro
where the upper limit is (ubl ) in the lth length, the lower limit is
plant are Vhj,min and Vhj,max respectively.
(ubl ) in the lth length, (T̂l ) is the amount of lth length the objective
(best solution discovered yet), and (c) is a reducing coefficient to
2.2.7. Balance of water dynamic decrease the size of all three areas: rest, aversion, and attraction.
The constraint related to the balance of water dynamic is given Clearly, from this equation, the swarm improves the location near
by Eq. (9): an objective (T̂l ). The feature (c) makes the convergence of swarm
Nj
gets an orientation to the objective. In the GOA algorithm, it
∑ t −λmj has estimated that the aim has been achieved the best solution
Vhjt = Vhjt −1 + t
Ihj − Qhjt + Qhm (9)
yet. While grasshoppers were interacting and keeping track of
m=1
the objective, the optimal solution was improved when a more
where the entire of upstream hydro plants straight overhead the desirable solution was found.
jth hydro plant is (Nj ). The feature (c) is the essential regulation feature in the GOA
algorithm and has been improved according to Eq. (14) (Aljarah
et al., 2018):
2.2.8. Initial and final reservoir storage ( )
The constraint related to the balance of water dynamic is given cmax − cmin
c = cmax − i (14)
by Eq. (10): I

Vhj0 = Vhj,ini ; VThj = Vhj,fin (10) where the maximum iterations number is (I), the existing it-
eration is (i), cmax is equal to 1, and cmin is equal to 0.00001.
where the initial and final reservoir storage of the jth hydro plant The aforementioned illustrates the objective’s location in a three-
is Vhj,ini and Vhj,fin respectively. dimensional space more than 1000 iterations of the algorithm.
316
X. Zeng, A.T. Hammid, N.M. Kumar et al. Energy Reports 7 (2021) 314–323

The mechanical implementation of the swarm motion near an of the specific unit of thermal production (i) can be determined
objective in a three-dimensional space by the athletic model. by using Eq. (18):
Owing to the engagement of vectors in its model, it can be an Ns Nh
increased number of dimensions. It is established that the GOA ∑ ∑
Psit = PDt − Psit + t
Phj ,
has been extremely useful in determining optimal solutions for
i=1 j=1
difficult problems. i̸ =d
Despite the easiness, the suggested algorithm correctly ap-
t∈T (18)
proaches solutions near the most suitable exploration extent. The
initial population was so significant in GOA due to the limita- The specific thermal production should be satisfied with the
tion of stochastic element numbers include the suggested algo- restrictions in Eq. (5). The function of cost F should be minimized
rithm. The GOA algorithm (excepting OGStHS in the suggested as shown in objective function Eq. (1).
approach) was of the number of solutions, iterations, and vari- For more clarification on the optimization process, the sug-
ables. For this aspect, the distance of one grasshopper to the gested GOA flow chart has been illustrated in Fig. 1. Besides, step-
others must be determined in every dimension throughout iter- wise procedure, i.e., a complete algorithm may then be expressed
ations. Moreover, arithmetic complication takes into account the in steps as shown below:
cost of the objective function, as illustrated in Eqs. (1) and (2) due
to the variation of this problem. Step-1. Indiscriminately initialize the grasshopper’s swarm pop-
ulation based on each unit’s limitations, such as different
4. Execution of GOA for optimal short-term hydrothermal lengths and exploring points. These initial grasshoppers
scheduling swarm should be possible to elect the final decision
that should be satisfied with grasshoppers to swarm
The implementation of the algorithm GOA method to solve operation constraints.
the OGStHS problem is explained in detail in this section. The Step-2. Made a comparison for every value of agent cost with
significant employment of the suggested approach is essentially that of its parameters includes Gi , Gr , and We . The agent
based on the mechanism that handles the constraints. Therefore, with the best cost value has been signified as the best
this section essentially concentrates on this subject. Based on interaction based on Eq. (11).
the properties of the OGStHS problem, it has been selected a Step-3. Adjust the motion of each agent based on Eq. (12).
collection of water release rates (Qhjt ) as the results variables for
Step-4. Adjust the interaction of each agent based on Eq. (13).
hydro plants while applying a collection of energy productions
Step-5. For each individual X-matrix in the population illus-
(Psit ) as the results variables for thermal plants. For a specific
trated in Eq. (15), it must be determined the function
period of time (T ) through the schedule extent, T will schedule
of cost F consistent with Eq. (1).
water releases rates by Nh hydro plants and T to the energy
Step-6. If the newcomer value of cost for any ith agent is less
production schedule by Ns thermal plants. Therefore, the solution
than its earlier rate, the agent’s newcomer arrangements
description executed in this article to solve the OGStHS problem
would be kept as its Gi . For finding the best Xil value, the
is as follows:
The scale of the components Psit and Qhjt must be satisfied cost values of Gi are compared for every agent.
with the capacity of thermal producing and the rate of water Step-7. If iterations number gets to the maximum, then turn to
release constraints according to Eqs. (5) and (7). Considering the the next step, else, return to adjust the interaction again.
spillage in Eq. (9) is zero to make easiness, the hydraulic sequence Step-8. For every individual that produces the most advanced,
constraints have to, as shown in Eqs. (15) and (16). which is regarded as Xil is made a problem solution.

Q1,0 Q2,0 ··· QN ,0 Ps1,0 Ps1,0 ··· PsN ,0

⎡ ⎤
5. Results and discussion
⎢Q1,1 Q2,1 ··· QN ,1 Ps1,1 Ps2,1 ··· PsN ,1 ⎥
⎣ ..
X =⎢ .. .. .. .. .. ⎥ (15)
The suggested algorithm is examined with two individual sys-
. . ··· . . . ··· .
⎦
Q1,T Q2,T ··· QN ,T Ps1,T Ps2,T ··· PsN ,T tems, the first system containing 9-buses, eleven-transmission-
lines, 8 thermal plants, 6 hydro plants, and the second with
T T Up T
an adjusted standard efficient system, including 65-buses 92-
T −tq transmission-lines, twelve thermal plants, and eleven hydro
∑ ∑ ∑ ∑
Vhj0 − VhjT = QhjT − Qhl − t
Ihj ,
plants. The suggested GOA is employed to make the solution
t =1 t =1 l=1 t =1
to the OGStHS problem. To prevent fake results because of the
j ∈ Nh (16) stochastic characteristics of the GOA, 1000 trial runs have been
Rate of water steam of jth reservoir at a specific period of time executed with various random populations at every run. The
t
t is (Ihj ), delay of a water passage from reservoir l to j is (tq ), extent of the population was 60-agents in the whole run. The
and upstream units numbers straight overhead jth hydro plant simulation has been performed on a core i7, 8th Gen with a
is (Up ). To match precisely the constraints on the initial and final 2.00 GHz processor.
reservoir storage in Eq. (9), the rate of water release of the jth From the aspect of convergence of the hydrothermal schedul-
hydro plant Qhjt at a specific period of time t is determined by ing of a fitness function for better five individuals (ind) than
Eq. (17): others of II system using suggested GOA algorithm has been
displayed in Fig. 2. The convergence feature of the fitness func-
T T Up T
∑ ∑ ∑ T − tq
∑ tion has been described by selecting the original value of mini-
Qhjt = Vhj0 − Vhjt + t
Ihj + Qhl − Qhjt , mum fitness when iterations reached to end. It can be observed
t =1 t =1 l=1 t =1
from Fig. 2 that convergence of fitness function gets to the opti-
t ̸ =d
mum value easily with no sudden oscillations. This displays the
j ∈ Nh (17)
accuracy of convergence for the suggested algorithm.
The specific rate of water release should be satisfied with Fig. 3 displays the daily load curve of the I system and the
the constraints in Eq. (7). Moreover, to match precisely the con- adjusted II system. Fig. 4 demonstrates the paths of the optimal
straints of power balance in Eq. (4), the thermal production (Psit ) release of the hydro plants of the I system. It may be noticed from
317
X. Zeng, A.T. Hammid, N.M. Kumar et al. Energy Reports 7 (2021) 314–323

Fig. 1. Schematic diagram of the proposed GOA-HGS model.

Fig. 4 that the paths of hydro release achieved by the suggested optimal appearance presented separately through the best non-
GOA exceedingly correspond with the daily load curve. dominated results in Q achieved through GOA has been nearer to
According to the I system, Fig. 5 displays the Pareto distribu- the real Pareto appearance than that of GA-2. While from Fig. 5,
tion for best final decisions in Q received by GOA and improved we may notice that the Pareto best results achieved by GOA are
generic algorithm type 2 (GA-2) (Lu et al., 2011). It regards as adequately distributed on the best appearance. Though the Pareto
an objective function value of gotten absolute decision meth- best results achieved by GA-2 are appropriately distributed, the
variegation characteristics are not as well as the ones archived by
ods. In comparison, all aspects of the generations for hydrother-
GOA.
mal power and water release rates of GOA’s optimal adjust-
For the II system, it can be noticed, according to Fig. 6, that
ment schedule solutions are delivered in Table 1. The reservoir
GOA can make a solution to this system with satisfying perfor-
storage volumes per hour of the optimal adjustment schedule mance while GA-2 cannot be. As displayed in Fig. 6, the Pareto
solutions have also been evaluated to verify whether the problem best results achieved by GOA are properly dispersed on the ap-
constraints have been satisfied or not to be. pearance with satisfying variety. Their convergence characteris-
From Fig. 5, we may also see the performance for handling tics are better than the solutions achieved through GA-2. Fur-
the suggested GOA system is extremely better than the solu- thermore, from Fig. 6, it can be assumed that GA-2 cannot deal
tions achieved through GA-2. According to Fig. 5, the Pareto with this type of system because the Pareto best results achieved
318
X. Zeng, A.T. Hammid, N.M. Kumar et al. Energy Reports 7 (2021) 314–323

Fig. 2. Convergence characteristics of II system.

Fig. 3. Load curve for I system and II system.

Fig. 4. Release paths of hydro plants – I system.

by it have been dispersed irregular on various Pareto appear- to confirm the suggested approach’s effectiveness. Based on Ta-
ances. Therefore, the suggested GOA approach’s convergence and ble 1 and Fig. 7, it is understood that the consideration of the
variegation characteristics for solving this type of system are transmission loss has grown the challenges of the problem sig-
adequately proved. nificantly. The restrictions processing plans originated in this
Simultaneously, with data arranged in Table 1 and displayed study can handle the complex restrictions of the OGStHS prob-
in Fig. 7, the rigid restriction requirements of the optimal adjust- lem efficiently. The final decisions of scheduling achieved by
ment schedule results achieved through GOA have been verified the suggested GOA approach can be satisfied entire types of
319
X. Zeng, A.T. Hammid, N.M. Kumar et al. Energy Reports 7 (2021) 314–323

Fig. 5. Pareto optimal front obtained by a different method for the I system.

Fig. 6. Various methods achieve Pareto best appearance.

Table 1
The aspects of the optimal compromise solution achieved by GOA.
Hour Water discharge rates × 104 m3 Hydro generations in MW Thermal generations in MW Total in MW
1 8.624 7.111 27.552 7.111 81.686 58.721 15.597 141.601 181.587 189.111 147.95 816.253
2 8.68 7.111 26.611 7.111 82.743 59.885 15.652 138.748 183.933 216.77 148.522 846.253
3 8.299 7.111 24.54 7.111 80.481 61.017 22.991 135.465 182.8 134.045 149.454 766.253
4 8.403 7.111 21.605 7.111 81.41 62.655 34.924 131.346 127.766 133.912 144.239 716.252
5 7.009 7.111 18.935 7.111 70.755 64.221 44.309 125.543 151.949 134.117 145.358 736.252
6 6.692 7.111 18.937 7.111 68.137 65.225 43.604 144.428 184.123 211.839 148.897 866.253
7 11.422 9.092 18.169 12.14 99.163 78.989 45.948 223.348 184.123 232.102 152.58 1016.253
8 11.378 9.194 17.699 16.887 98.204 78.569 46.448 280.713 184.123 224.769 163.428 1076.254
9 11.144 9.189 19.017 18.793 96.709 77.955 40.815 300.196 184.123 218.939 237.516 1156.253
10 10.398 8.947 19.697 18.937 92.997 76.443 37.884 301.385 184.096 218.997 234.45 1146.252
11 11.18 9.704 18.788 18.169 97.436 81.562 41.676 295.62 184.123 227.706 238.13 1166.253
12 10.788 9.397 19.392 17.699 96.139 79.999 40.063 291.914 184.123 301.208 222.807 1216.253
13 10.701 9.625 19.642 19.017 95.793 81.174 39.557 301.963 184.123 236.887 236.757 1176.254
14 10.592 10.241 19.892 19.697 95.631 84.322 40.034 306.733 184.123 233.177 152.232 1096.252
15 10.384 10.18 19.067 18.788 95.176 83.921 44.308 300.295 184.123 219.63 148.8 1076.253
16 10.293 10.506 18.178 19.392 95.091 85.619 48.464 304.628 184.123 238.034 170.293 1126.252
17 10.393 10.996 17.299 19.642 95.846 87.307 52.025 306.354 184.123 224.975 165.625 1116.255
18 10.244 11.544 16.855 19.892 94.936 88.187 54.602 308.044 184.123 221.456 234.905 1186.253
19 10.286 12.181 16.467 21.103 94.91 88.139 56.948 315.674 184.123 223.546 172.913 1136.253
20 10.262 12.997 16.176 21.101 94.233 88.5 58.877 313.575 184.123 220.42 156.525 1116.253
21 6.15 10.715 12.752 20.137 63.689 76.239 62.197 304.658 184.123 136.109 149.238 976.253
22 6.121 11.44 13.137 20.138 63.72 79.165 64.805 301.736 136.438 134.036 146.354 926.254
23 6.111 11.863 13.899 20.795 64.023 80.012 67.243 302.242 121.173 134.031 147.529 916.253
24 6.111 8.185 13.506 20.766 64.426 58.903 68.156 297.355 113.899 134.31 129.204 866.253

this system constraints. Therefore, the GOA approach’s successful Moreover, the optimal power production and rate of reservoir
employment to solve the OGStHS problem is adequately proved release for only I system have been displayed in Fig. 8, which
again. Simultaneously, the productions of hydrothermal power regards statistical analysis. It was evident from Fig. 8 that the sug-
and rates of water release of the optimal adjustment schedule gested method provides practical and best results. To verify the
solution achieved by GOA have been arranged in Table 1, and its efficiency of the suggested GOA, it has been made a comparison
reservoir storage volumes per hour have been displayed in Fig. 7 with the Decomposition Approach (DA) and the LP method and
to verify the contentment conditions to constraints. the GA as described in the previously completed works section of
320
X. Zeng, A.T. Hammid, N.M. Kumar et al. Energy Reports 7 (2021) 314–323

Fig. 7. Scheme of hourly hydro reservoir volumes of the optimal adjustment scheduling achieved by GOA.

Fig. 8. Optimal power generation achieved by GOA technique.

Table 2
Comparison of cost and time of a 66-bus utility system (For I system).
Method Time (s) Cost ($) Processor
Min Mean Max
DP (Lakshminarasimman and Subramanian, 2006) – 928,919.15 – – C++ code on a Pentium-IV
DE (Wang et al., 2012b) 8.69 923,991.08 925,157.28 928,395.84 Delphi 2010 on a P-IV
IPSO (Hota et al., 2009) 38.46 922,553.49 – – Matlab-7.0 on a P-IV
DRQEA (Wang et al., 2012b) 7.98 922,526.73 923,419.37 925,871.51 Delphi 2010 on a P-IV
CRQEA (Wang et al., 2012a) – 922,477.14 – – Delphi 2010 on a P-IV
MAPSO (Amjady and Soleymanpour, 2010) 64 922421.66 922,544 923,508 –
TLBO (Roy, 2013) – 922,373.39 922,462.24 922,873.81 Core 2 duo
RCGA-AFSA (Carvalho and Soares, 1987) 11 922,339.625 922,346.323 922,362.532 Microcomputer (64 kB)
SPPSO (Zhang et al., 2011) 16.3 922,336.31 923,083.48 922,362.532 C++ code in the Linux (Core 2 Duo)
DNLPSO (Rasoulzadeh-Akhijahani and Mohammadi-Ivatloo, 2015) 37 922,498 922,837 923,580 –
MDNLPSO (Rasoulzadeh-Akhijahani and Mohammadi-Ivatloo, 2015) 35 922,336.3 922,676.2 923,404.5 –
SOS (Das and Bhattacharya, 2018) 6.21 922,332.1691 922,338.1982 922,482.8956 MATLAB-2013 on a core i3
GA (Kumar and Mohan, 2011) 4.9 – – 78,757.12 Pentium(R)
DA and LP method (Mohan et al., 1992) 56.2 – – 78,654.86 WIPRO-386
Proposed GOA 3.8 78,650 78,653.56 78,657.12 Core i7-8th

introduction (Mohan et al., 1992) and Kumar and Mohan (2011) studies to solve the SHS problem using various modern tech-
respectively. niques. In this comparison, the focused methods include the
The I system and II system data from the mentioned section DP, DE, improved PSO (IPSO), differential real-coded quantum-
are taken into account for making this performance differenti- inspired evolutionary algorithm (DRQEA), modified adaptive PSO
ation. Table 2 displays the differentiation of cost and achieve- (MAPSO), hybrid of real coded genetic algorithm and artificial fish
ment period for the I system. Furthermore, this section has been swarm algorithm (RCGA-AFSA), small population PSO (SPPSO),
used to compare the present study results with other major dynamic neighborhood learning PSO (DNLPSO), modified DNLPSO
321
X. Zeng, A.T. Hammid, N.M. Kumar et al. Energy Reports 7 (2021) 314–323

Table 3
Comparison of cost and time of a 66-bus utility system (For II system).
Method Time (s) Cost ($)
Min Mean Max
TLBO (Roy, 2013) – – – 423,858.78
PSO (Das and Bhattacharya, 2018) 125.32 318,970 319,350.84 320,874.21
SOS (Das and Bhattacharya, 2018) 46.25 314,994.38 315,052.3 315,718.42
GA (Kumar and Mohan, 2011) 12.765 – – 226,205.76
DA and LP method (Mohan et al., 1992) 89.4 – – 220,537.46
Proposed GOA 11.654 219,899.24 220,025.5 220,205.76

Table 4 improve the searching ability of the used algorithm. As a result,

p values of Wilcoxon signed-rank test result for GOA algorithm. GOA decreases the difficulty and computation time and provides
GOA versus GA-2 GOA versus GA a globally optimum solution. We have made a comparison to
I System 8.2021E−10 10.5354E−10 the DA, LP, and GA method. Moreover, the suggested method’s
II System 9.9757E−10 11.7576E−10 robustness has been made a validation using executing the non-
parametric Wilcoxon signed-rank analysis. Overall, it is realized
that the GOA approach may reach more reliable solutions for the
(MDNLPSO), symbiotic organisms search (SOS), DA, LP, and GA OGStHS with robust performance in less computational time than
technique. the recent techniques.
It has been found out that the cost of the suggested GOA is
relatively similar to that of the DA, LP, and GA technique, and CRediT authorship contribution statement
the achievement time of the suggested GOA is significantly less
when we made a comparison with that of the DA, LP, and GA Xie Zeng: Conceptualization, Methodology, Formal analysis,
technique. Thus, GOA’s maximum and the average cost is better Investigation, Data curation, Writing - original draft, Funding ac-
than that of other methods described in the above section. In quisition. Ali Thaeer Hammid: Conceptualization, Methodology,
the same regard, Table 3 displays the differentiation of cost and Formal analysis, Investigation, Data curation, Writing - origi-
achievement period for the II system used. For comparing system nal draft. Nallapaneni Manoj Kumar: Conceptualization, Formal
11, TLBO, PSO, SOS, GA, DA, and LP methods were used. analysis, Supervision, Visualization, Writing - review & editing.
Umashankar Subramaniam: Writing - review & editing, Funding
Nevertheless, Table 3 (II system) has got the same range re-
acquisition. Dhafer J. Almakhles: Writing - review & editing,
sults as Table 2 (I System). The GOA is better after the comparison
Funding acquisition.
of the cost values. These were relatively similar to that of the DA,
LP, and GA technique, and the achievement time of the suggested
Declaration of competing interest
GOA is significantly less. Thus, GOA’s maximum and the average
cost is better than that of other methods described in Tables 2
The authors declare that they have no known competing finan-
and 3.
cial interests or personal relationships that could have appeared
To validate the analysis’ performance statistically, forty confi-
to influence the work reported in this paper.
dent trials have been provided to GA, improved GA-2, and GOA.
In contrast, the whole selection features of algorithms have been
Acknowledgments
established to be corresponding to the OGStHS analysis system.
The cost and time of achieved solutions by DA, LP, GA, and GOA The authors would like to acknowledge the support of Prince
techniques have been displayed in Table 4, and it was evident Sultan University for paying the Article Processing Charges (APC)
from these solutions that GOA outperforms other methods. of this publication. Also, the first author acknowledges the tech-
Therefore, it is shown that the GOA method has a very less nical support provided for him under the Key Research and De-
sensitive effect on the initial results. In this study, to compare the velopment Program in Shaanxi Province for the year 2020.
performance of GOA with GA and GA-2, nonparametric Wilcoxon
signed-rank analysis has been employed in the analysis system. Funding
The Wilcoxon signed-rank analysis support to distinguish the
critical variation between two individual means (Derrac et al., This work was supported by the Renewable Energy Lab, Col-
2011). The investigation has been completed at a level of impor- lege of Engineering, Prince Sultan University, Saudi Arabia.
tance α = 0.01. Results analysis is listed in Table 4, and it has
been recognized that there is an important variance between GOA References
in comparison with GA and GA-2 method.
Aljarah, I., et al., 2018. Simultaneous feature selection and support vector
6. Conclusion machine optimization using the grasshopper optimization algorithm. Cogn.
Comput. 10 (3), 478–495.
Amjady, N., Soleymanpour, H.R., 2010. Daily hydrothermal generation scheduling
The OGStHS has been a complex arithmetic optimization prob- by a new modified adaptive particle swarm optimization technique. Electr.
lem with extremely nonlinear and calculational exorbitant con- Power Syst. Res. 80 (6), 723–732.
ditions. The study presented a method to solve this problem by Basu, M., 2004. An interactive fuzzy satisfying method based on evolution-
ary programming technique for multiobjective short-term hydrothermal
the GOA method in which whole variables have been considered scheduling. Electr. Power Syst. Res. 69 (2–3), 277–285.
without obtaining the common facilitate presumptions needed Carvalho, M., Soares, S., 1987. An efficient hydrothermal scheduling algorithm.
by traditional methods. The suggested algorithm GOA has been IEEE Trans. Power Syst. 2 (3), 537–542.
examined on two individual systems. The GOA system displays Chang, S.-C., et al., 1990. Hydroelectric generation scheduling with an effective
differential dynamic programming algorithm. IEEE Trans. Power Syst. 5 (3),
the high ability to explore the optimal generation schedule due to 737–743.
the main factor: group interaction of employment swarms. More- Das, S., Bhattacharya, A., 2018. Symbiotic organisms search algorithm for
over, each other factors (gravity and wind effect horizontally) short-term hydrothermal scheduling. Ain Shams Eng. J. 9 (4), 499–516.

322
X. Zeng, A.T. Hammid, N.M. Kumar et al. Energy Reports 7 (2021) 314–323

Das, S., Bhattacharya, A., Chakraborty, A.K., 2018. Fixed head short-term hy- Nguyen, T.T., Vo, D.N., Dinh, B.H., 2018. An effectively adaptive selective cuckoo
drothermal scheduling in presence of solar and wind power. Energy Strategy search algorithm for solving three complicated short-term hydrothermal
Rev. 22, 47–60. scheduling problems. Energy 155, 930–956.
Derrac, J., et al., 2011. A practical tutorial on the use of nonparametric statistical Orero, S., Irving, M., 1998. A genetic algorithm modelling framework and solution
tests as a methodology for comparing evolutionary and swarm intelligence technique for short term optimal hydrothermal scheduling. IEEE Trans.
algorithms. Swarm Evol. Comput. 1 (1), 3–18. Power Syst. 13 (2), 501–518.
Guan, X., Luh, P.B., Zhang, L., 1995. Nonlinear approximation method in Piekutowski, M.R., Litwinowicz, T., Frowd, R., 1993. Optimal short-term schedul-
Lagrangian relaxation-based algorithms for hydrothermal scheduling. IEEE ing for a large-scale cascaded hydro system. In: Conference Proceedings
Trans. Power Syst. 10 (2), 772–778. Power Industry Computer Application Conference. IEEE.
Hammid, A.T., Sulaiman, M.H.B., 2018. Series division method based on PSO Rasoulzadeh-Akhijahani, A., Mohammadi-Ivatloo, B., 2015. Short-term hydrother-
and FA to optimize long-term hydro generation scheduling. Sustain. Energy mal generation scheduling by a modified dynamic neighborhood learning
Technol. Assess. 29, 106–118. based particle swarm optimization. Int. J. Electr. Power Energy Syst. 67,
Hammid, A.T., Sulaiman, M.H.B., Abdalla, A.N., 2018a. Prediction of small hy- 350–367.
dropower plant power production in himreen lake dam (HLD) using artificial Roy, P.K., 2013. Teaching learning based optimization for short-term hydrother-
neural network. Alexandria Eng. J. 57 (1), 211–221. mal scheduling problem considering valve point effect and prohibited
Hammid, A.T., Sulaiman, M.H.B., Awad, O.I., 2018b. A robust firefly algo- discharge constraint. Int. J. Electr. Power Energy Syst. 53, 10–19.
rithm with backpropagation neural networks for solving hydrogeneration Salam, M.S., Nor, K.M., Hamdam, A., 1998. Hydrothermal scheduling based
prediction. Electr. Eng. 100 (4), 2617–2633. Lagrangian relaxation approach to hydrothermal coordination. IEEE Trans.
Hammid, A., et al., 2017. Priority of kaplan turbine and small hydropower plants Power Syst. 13 (1), 226–235.
over other resources: An overview. J. Eng. Appl. Sci. 12 (9), 8695–8705. Saremi, S., Mirjalili, S., Lewis, A., 2017. Grasshopper optimisation algorithm:
Hota, P., Barisal, A., Chakrabarti, R., 2009. An improved PSO technique for theory and application. Adv. Eng. Softw. 105, 30–47.
short-term optimal hydrothermal scheduling. Electr. Power Syst. Res. 79 (7), Sharma, K., Dubey, H.M., Pandit, M., 2020. Short-term hydrothermal scheduling
1047–1053. using gray wolf optimization. In: Advances in Computing and Intelligent
Jian, J., Pan, S., Yang, L., 2019. Solution for short-term hydrothermal scheduling Systems. Springer, pp. 253–269.
with a logarithmic size mixed-integer linear programming formulation. Sinha, N., Chakrabarti, R., Chattopadhyay, P., 2003. Fast evolutionary program-
Energy 171, 770–784. ming techniques for short-term hydrothermal scheduling. IEEE Trans. Power
Kumar, V.S., Mohan, M., 2011. A genetic algorithm solution to the optimal Syst. 18 (1), 214–220.
short-term hydrothermal scheduling. Int. J. Electr. Power Energy Syst. 33 Tang, J., Luh, P.B., 1995. Hydrothermal scheduling via extended differential
(4), 827–835. dynamic programming and mixed coordination. IEEE Trans. Power Syst. 10
Lakshminarasimman, L., Subramanian, S., 2006. Short-term scheduling of hy- (4), 2021–2028.
drothermal power system with cascaded reservoirs by using modified Tehzeeb ul, H., et al., 2020. Short-term optimal scheduling of hydrothermal
differential evolution. IEE Proc., Gener. Transm. Distrib. 153 (6), 693–700. power plants using artificial bee colony algorithm. Energy Rep. 6, 984–992.
Lakshminarasimman, L., Subramanian, S., 2008. A modified hybrid differential Turgeon, A., 1981. Optimal short-term hydro scheduling from the principle of
evolution for short-term scheduling of hydrothermal power systems with progressive optimality. Water Resour. Res. 17 (3), 481–486.
cascaded reservoirs. Energy Convers. Manage. 49 (10), 2513–2521. Wang, Y., et al., 2012a. A clonal real-coded quantum-inspired evolutionary
Lu, Y., et al., 2010. An adaptive chaotic differential evolution for the short-term algorithm with Cauchy mutation for short-term hydrothermal generation
hydrothermal generation scheduling problem. Energy Convers. Manage. 51 scheduling. Int. J. Electr. Power Energy Syst. 43 (1), 1228–1240.
(7), 1481–1490. Wang, Y., et al., 2012b. Short-term hydrothermal generation scheduling using
Lu, Y., et al., 2011. A hybrid multi-objective cultural algorithm for short-term differential real-coded quantum-inspired evolutionary algorithm. Energy 44
environmental/economic hydrothermal scheduling. Energy Convers. Manage. (1), 657–671.
52 (5), 2121–2134. Wolpert, D.H., Macready, W.G., 1997. No free lunch theorems for optimization.
Mandal, K.K., Basu, M., Chakraborty, N., 2008. Particle swarm optimization IEEE Trans. Evol. Comput. 1 (1), 67–82.
technique based short-term hydrothermal scheduling. Appl. Soft Comput. 8 Wong, K., Wong, Y., 1994. Short-term hydrothermal scheduling part. I. Simulated
(4), 1392–1399. annealing approach. IEE Proc., Gener. Transm. Distrib. 141 (5), 497–501.
Mandal, K., Chakraborty, N., 2008. Differential evolution technique-based short- Yuan, X., Yuan, Y., Zhang, Y., 2002. A hybrid chaotic genetic algorithm for
term economic generation scheduling of hydrothermal systems. Electr. Power short-term hydro system scheduling. Math. Comput. Simulation 59 (4),
Syst. Res. 78 (11), 1972–1979. 319–327.
Mohan, M.R., Kuppusamy, K., Khan, M.A., 1992. Optimal short-term hydrothermal Zhang, J., Wang, J., Yue, C., 2011. Small population-based particle swarm
scheduling using decomposition approach and linear programming method. optimization for short-term hydrothermal scheduling. IEEE Trans. Power
Int. J. Electr. Power Energy Syst. 14 (1), 39–44. Syst. 27 (1), 142–152.
Nazari-Heris, M., et al., 2018. Improved harmony search algorithm for the solu- Zhang, J., et al., 2016. A hybrid particle swarm optimization with small popu-
tion of non-linear non-convex short-term hydrothermal scheduling. Energy lation size to solve the optimal short-term hydrothermal unit commitment
151, 226–237. problem. Energy 109, 765–780.

323