Optimal Power Flow Solution in the

Presence of Renewable Energy Sources

Mahaboob Shareef Syed, Suresh

V. Chintalapudi & Sivanagaraju Sirigiri

Iranian Journal of Science and

Technology, Transactions of
Electrical Engineering

ISSN 2228-6179
Volume 45
Number 1

Iran J Sci Technol Trans Electr Eng

(2021) 45:61-79
DOI 10.1007/s40998-020-00339-z

1 23
Your article is protected by copyright and
all rights are held exclusively by Shiraz
University. This e-offprint is for personal
use only and shall not be self-archived
in electronic repositories. If you wish to
self-archive your article, please use the
accepted manuscript version for posting on
your own website. You may further deposit
the accepted manuscript version in any
repository, provided it is only made publicly
available 12 months after official publication
or later and provided acknowledgement is
given to the original source of publication
and a link is inserted to the published article
on Springer's website. The link must be
accompanied by the following text: "The final
publication is available at link.springer.com”.

1 23
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79
https://doi.org/10.1007/s40998-020-00339-z (0123456789().,-volV)(0123456789().,-volV)

RESEARCH PAPER

Optimal Power Flow Solution in the Presence of Renewable Energy

Sources
Mahaboob Shareef Syed1 • Suresh V. Chintalapudi2 • Sivanagaraju Sirigiri3

Received: 18 July 2019 / Accepted: 21 April 2020 / Published online: 28 May 2020
 Shiraz University 2020

Abstract
In the present work, the renewable energy sources are integrated into the power system. For this purpose, the wind-rich
locations and sunny areas are identified in the considered system. A novel persistence-extreme learning machine algorithm
is proposed. The wind speed and solar insolations are forecasted in short-term and long-term time periods in the identified
areas using the proposed method. With the penetration of wind and solar powers in the system, the optimal power flow
problem is solved in 12 different identified cases. The results are analyzed in the prospective of voltage deviation and
active power loss. It is observed that the voltage deviation is more in the short-term as well as long-term time horizons with
wind and solar integration, but the active power losses are less compared to remaining cases. The analysis is carried out by
considering Andhra Pradesh—14-bus system and a 124-bus Indian utility real time system (IND-124). The entire simu-
lation is carried out with the help of MATLAB 2013a.

Keywords Forecasting  Persistant-Extreme learning machine algorithm  Renewable energy sources  Renewable
integration  Optimal power flow solution

1 Introduction characterized by their intermittent nature of power gener-

ation. When these interconnect to the grid along with the
Optimal power flow (OPF) is a highly nonlinear complex conventional generators, the operation of power system
problem that simultaneously requires vital analysis of remains complicated. Khan and Singh (2017), the authors
power system planning, secure operation and control. It presented the rigorous, comprehensive review of the OPF
was first introduced by Wood and Wollenberg (2013). This problem in its different forms. OPF with conventional
problem is to combine power flow with the economic sources as well as RES integration, storage systems, by
dispatch, i.e., solving for optimum generation while enforcing different system constraints and solving with
imposing the limits on the transmission system (Duncan various techniques and methodologies, is presented.
Glover et al. 2016) to obtain secure operation. RESs are Shilaja and Ravi (2016), the OPF problem with and
without wind generation is studied using the PSO algo-
rithm. The uncertainty of wind generation is handled with
& Mahaboob Shareef Syed
[email protected] Weibull distribution function. Biswas et al. (2017), the
OPF problem with the integration of wind and solar in
Suresh V. Chintalapudi
[email protected] IEEE 30-bus system is solved using success-history-based
parameter adaptation technique of differential evolution
Sivanagaraju Sirigiri
[email protected] combined with superiority of feasible solutions (SHADE-
SF). The OPF is studied in different test conditions by
1
EEE Department, VLITS, Vadlamudi, Guntur Dt, modeling wind and solar generations with different prob-
AP 522213, India ability distribution functions.
2
Department of EEE, VVIT, Nambur, Guntur Dt, AP 522508, Kathiravan and Kumudini Devi (2017), classical OPF is
India solved with the integration of solar, wind and bundled
3
EEE Department, UCEK, JNTUK, Kakinada, E.G. Dt, solar-thermal power technologies in various test cases
AP 533003, India

123
 Author's personal copy
62 Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79

using flower pollination algorithm. Surender Reddy (2017), the system, it demands the use of sophisticated forecasting
the OPF problem is solved by the RES including storage tool. Lahari and Vasantha Kumar Shetty (2015) and Car-
system. A new strategy is developed by authors to handle denas et al. (2017), a major literature survey has been
the uncertain nature of the load, wind and solar generation, carried out for wind speed or wind power forecasting over
and the performance is evaluated in comparison with GA, a different time horizon. Similarly, Peng et al. (2017) and
two-point estimation method and Monte Carlo simulation Lotfy et al. (2018) has been focused on solar power
(MCS). forecasting.
A security constrained OPF (SCOPF) problem is solved in The state-of-the-art forecasting puts forward the meth-
Khazal and Kalantar (2015) with wind penetration as well as ods, which are persistence, physical, statistical, artificial
responsive demands. Beta distribution and Gaussian distri- neural network and finally the hybrid models. Persistence is
bution are used for modeling wind power and load, respec- a naive and simple method of forecasting, used for a short
tively. The authors consider operating costs, security indices period time horizons. Physical forecasting methods make
and emission as single as well as multi-objective functions. use of physical atmospheric parameters such as wind
Teeparthi and Vinod Kumar (2017), a security constraint speed, pressure and temperature. Statistical methods make
wind integrated multi-objective OPF is solved with the use of time series and regressive approaches. The artificial
help of hybrid particle swarm optimization and artificial neural network needs historical data, and finally by com-
physics optimization (HPSO–APO). The objectives include bining above-mentioned parameters, any two hybrid
active power loss, total production cost and security indices approach methods are developed.
for base case as well as in contingency scenario. A SCOPF In this present effort, the wind speed and solar insolation
is solved in Teeparthi and Vinod Kumar (2018) using fuzzy have been forecasted using a P-ELM algorithm, which is a
adaptive-based artificial physics optimization algorithm combination of both persistence and extreme learning
(FAAPO) with wind integration. The wind is modeled by machine (ELM) algorithms (Sekar and Duraisamy 2015).
Weibull distribution function, and the problem is solved by The persistence approach assumes that the forecasted value
considering penalty and reserve costs both in normal and at (t ? 1)th instant is the same as that of at the tth instant,
contingency cases. which is most efficient for shorter time horizons. ELM
Alasan et al. (2019), for nonlinear programming a novel algorithm is a single hidden layer feed-forward neural
CONOPT solver is adopted which is embedded in the network.
package of GAMS in the process of OPF solution. The The mathematical modeling of P-ELM is demonstrated
uncertainty in load and renewable generation are handled as follows:
in Chamanbaz et al. (2017) by defining uncertainty prob- Let the samples of the P-ELM structure be ðsi ; siþ1 ÞN1 i¼1 ,
ability distribution aimed at optimizing the conventional si 2 RN , where si ¼ ½si1 ; si2; si3 ; . . .sip  characterizes the
generation. input samples and siþ1 ¼ ½sðiþ1Þ1 ; sðiþ1Þ2 ; sðiþ1Þ3 ; . . .; sðiþ1Þp 
From the above literature, it is identified that the existing are their respective output sample and p indicates number
literature is confined to study the control characteristics and of attributes in an individual sample. The mathematical
grid connection problems with RES. Further, it is required modeling of P-ELM with f (.) activation function, which is
to analyze the effect of RES on the system performance. In to be minimized, can be expressed as
this, a mathematical model to solve optimal power flow
problem with RES is presented. The analysis is performed X
M
Fðsj Þ ¼ ci f ðai  sj þ bi Þ; j ¼ 1; 2; 3. . .N ð1Þ
for the considered time horizons in different seasons for i¼1
short-term and long-term periods.
The novelty of the present work is to propose a novel where M is the number of hidden layer nodes, c is the
hybrid machine learning algorithm: P-ELM to forecast weight matrix connecting hidden layer and output layer, a
wind and solar powers, to integrate into the power grid and is the weight matrix between input layer and hidden layer,
to analyze the effects in terms of techno-economic aspects. and b is a threshold given to each neuron in the hidden
The simulation is carried out with the help of MATLAB layer. The total structure of P-ELM is shown in Fig. 1.
2013a package. In matrix form
f c¼S ð2Þ
S is the target matrix, whose values are:
2 Proposed P-ELM Method for Forecasting
S ¼ ½s2 ; s3 ; s4 ; . . .sN 
It is very essential to forecast renewable energy generation
The activation function f (.) can be represented as
accurately and precisely to integrate them into the central
electric grid. For proper planning, operation and security of

123
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79 63

b1
where ‘q’ is air mass density, ‘A’ is the blade swept area,
ai1
c1(i+1) ‘l’ is the length of the blade, and ‘r’ is the radius of area
ai2 b2 swept.
Input Parameters

c2(i+1)

Output Parameters
The wind turbine blades swept area is perfectly shown in
.
si+1th Fig. 2.
sith sample b3

.
.
sample
Upon substituting Eq. (8) in Eq. (6), we get

.
.

dl

.
PWind ¼ 0:5  m2  q  A  ð9Þ
cM(i+1) dt
.
aiM
f (a*s+b) PWind ¼ 0:5  q  A  m3 ð10Þ
bM
Input Layer Hidden Layer Output Layer According to Betz law, the maximum mechanical
energy that can be extracted from the kinetic energy is not
Fig. 1 Pictorial representation of P-ELM algorithm more than 59.3% irrespective of design. This is known as
Betz constant (CP) or power coefficient, and the
extractable wind power can be modified as
f ¼
2
f ða1  s1 þ b1 Þ f ða2  s1 þ b2 Þ : : f ðaM  s1 þ bM Þ
3 PWind ¼ 0:5  q  A  m3  CP ð11Þ
6 7
6 : 7 It can be observed from Eq. (11) that the wind power
6 7
6 : 7
6
6
7
7
generated is varied as the cube of wind velocity. From
4 : 5 this relation, a typical power curve of a wind turbine is
f ða1  sN1 þ b1 Þ f ða2  sN1 þ b2 Þ : : f ðaM  sN1 þ bM Þ N1M shown in Fig. 3. It is a curve drawn between wind
ð3Þ velocity in m/s and active power generated in p.u (Ack-
ermann 2012).
a and b attributes are selected arbitrarily. The cost function
The wind power generation can be explained in four
in back propagation algorithm is to minimize
zones.
" #2
N 1 X
X M
FBP ¼ ci f ðai  sj þ bi Þ  sjþ1 ð4Þ • In Zone A, the wind power generated is zero, as the
j¼1 i¼1 wind speed is less than the cut-in speed (VCut in). Cut-in
speed is the speed at which the turbine starts to generate
the power.
• Zone B lies in between cut-in speed and rated speed
3 Wind Power Calculation (Vrated) in which the power developed will increase
rapidly. At Vrated, the power developed will reach to
In general, the wind power will be produced by wind tur- rated output power capacity.
bines or windmills. The kinetic energy that exists in the • Zone C is in between Vrated and cut-out speed (VCut-Off
moving air will be converted to mechanical energy by the or VCut-out), in which constant output power is attained.
rotation of wind turbine blades, so electrical energy by the VCut-Off is the maximum allowable speed for power
electric generator set up. generation.
The kinetic energy stored in air particles of mass m
moving with VW velocity can be expressed as
1
Kinetic Energy ðKEW Þ ¼ m m2 ð5Þ
2
The rate of change of kinetic energy is expressed as
power available in the wind
d ðKEW Þ dm
PWind ¼ ¼ 0:5  m2  ð6Þ
dt dt
dm
But the rate of flow of mass dt is given as
dm dðq  VolumeÞ
¼ ð7Þ
dt dt Fig. 2 The area swept by the wind turbine blade
dl
¼ qA ð8Þ
dt

123
 Author's personal copy
64 Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79

Fig. 3 Variation of output wind power

• Beyond VCut-Off speed, which is Zone D, the turbine will

be stalled to avoid damage to the structure because of
heavy winds. From the above detailed discussion, the
wind power generation equation can be written as

PWind ¼ 0; m\VCut in
¼ 0:5  q  A  m3 ; VCut in \m\VCut out ð12Þ
¼ Pr ; m [ VCut out

Fig. 4 Equivalent circuit of a diode model of solar cell

4 Solar Power Generation

Solar energy is natural and freely available in abundant

nature. The main advantages of solar energy are: It is Coulomb; ‘VD’ = voltage across the diode in volts;
pollution free, requires less maintenance and has a long ‘K’ = Boltzmann constant which is 1.38*10-23;
life. The reason for the production of solar energy is the ‘T’ = temperature of cell in Kelvin; ‘Rsh’ = shunt resis-
radiation of the sun. It is a solar panel which is used to tance in Ohms.
convert solar insolation into electrical energy. A solar cell The output power equation of a solar cell is given as
or a PV cell is the fundamental component which forms PPV ¼ VPV  IPV ð14Þ
solar panel or solar arrays by connecting them in series and
When the solar insolation is known, the output power of
parallel combinations depending upon the requirement of
a solar plant (PS) in kW can be calculated as
voltage, current and power.
A typical solar cell consists of a semi-conductor P–N PS ¼ PR  GHI  AS  gPV ð15Þ
junction diode, which consists of a PN junction (Patel
where ‘PR’ is performance ratio, whose value lies in
2005). It works on the principle of the photovoltaic effect.
between 0.5 and 0.9, the default value is 0.75, ‘GHI’ is
When the sun radiation falls on the diode, the junction
global horizontal irradiance, i.e., solar insolation in kW/m2,
absorbs light and produce current. The equivalent circuit of
‘AS’ is area of solar panel in m2. ‘gPV ’is efficiency of a
a single diode model of solar cell (Khaligh and Onar 2017)
solar panel, its value is around 0.15.
consists of a current source connected to a diode which is
in forward bias as shown in Fig. 4.
‘Iph’ is photo-current and ‘Ish’ is shunt current. ‘Ipv’ is
5 Optimum location of RES Placement
solar output current in Amps which is given as
qVD VD The AP 14-bus system has five generator buses including
Ipv ¼ Iph  I0 eð KT Þ  ð13Þ
Rsh slack bus and nine load buses with 20 transmission lines.
where ‘I00 = reverse saturation current of diode in Amps; The bus data and line data are presented in ‘‘Appendix’’.
‘q’ = charge of electron whose value is 1.609 9 10-19 With the help of the data provided by New and Renewable

123
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79 65

18 same time using clearness index method (Syed et al. 2017),

the areas Anantapur, Kurnool, Kadapa and Nellore are
19 VIJAYANAGARAM(14)
identified as highest sunny areas over the remaining. The
20 single-line diagram of AP 14-bus system with the inte-
17 CHITTOOR(10) gration of RES is shown in Fig. 5.
KURNOOL(11) SRIKAKULAM(13) Similarly, IND-124 (Sunil Kumar et al. 2012) system
16
ANANTHAPUR(12) also is considered which has 28 generator buses including
PRAKASAM(9) one slack bus and 96 load buses with 237 transmission
11 12
13 15 lines. According to the data obtained from NREDCAP, the
9 areas Tadiparthi, Mahadevamangalam, Somayajulapalli
KADAPA(6) 14 and Rajampet are identified as wind-rich locations. And the
VISAKHAPATNAM(8)
areas Gooty, Dichpally Manubolu and Tandur are identi-
10 G EAST GODAVARI(7)
7 fied as the sunniest areas. The single-line diagram of IND
8 G
GUNTUR(5) 124-bus system is shown in Fig. 6.
WEST 6 To train the P-ELM network, the historical data are
GODAVARI(4) Nellore(3)
obtained from Indian solar resource data website. With the
5 4
P-ELM algorithm, the available wind and solar powers are
3 G
2 estimated in the identified areas. The RESs are integrated
1 KRISHNA(2) into the grid; the load flow and OPF problems are executed.
G
Slack (1)
G
6 Optimal Power Flow Solution
Fig. 5 Single line diagram of AP 14-bus system with RES with the Incorporation of RES

Energy Development Corporation of Andhra Pradesh The intension of solving OPF is to obtain the optimum
(NREDCAP), the areas Kurnool, Kadapa, Anantapur and control parameters to achieve a set of objectives of power
Chittoor are identified as the windiest locations. At the system. Usually, the objective considered is to re-dispatch

Fig. 6 Single line diagram of IND 124-bus system

123
 Author's personal copy
66 Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79

the generation level to obtain minimum operating cost where ‘CðPS;i Þ’ is the cost of ith solar generator to generate
while satisfying the system and operational constraints. PS MW, which is expressed in Eq. (24):
Mathematically, it is required to minimize the total cost
CðPS;i Þ ¼ KS  PS;i ð24Þ
(TC) which includes fuel cost of thermal plants, the cost of
wind generation and the cost of solar generation. The cost of under estimation (CSU,i) and cost of over
TC ¼ CT þ CW þ CS $=h ð16Þ estimation (CSO,i) are expressed as in Eqs. (25) and (26):
CSU;i ¼ KRS;i  ðPS Avail;i  PS;i Þ ð25Þ
The thermal fuel cost with N number of thermal plants is
given by CSO;i ¼ KPS;i  ðPS;i  PS Avail;i Þ ð26Þ
X
N
where ‘KS’ is a direct cost coefficient; ‘KRS,i’ is a reserve
CT ¼ CðPi Þ $=h ð17Þ cost coefficient for under estimation; ‘KPS,i’ is a penalty
i¼1
cost coefficients for overestimation of ith solar generator.
where ‘C(Pi)’ represents the cost function of ith thermal The objective function is solved subjected to the fol-
generator to generate Pi MW, which is given in its quad- lowing constraints:
ratic form as
a. Equality constraints
CðPi Þ ¼ ai  P2i þ bi  Pi þ ci $=h ð18Þ 1. Active power balance
where ‘ai’, ‘bi’ and ‘ci’ are the cost coefficients of ith The total active power generation must be equal to the
thermal generator. sum of system demand and transmission losses.
The cost of wind generator (CW) includes cost to gen- Mathematically,
erate PWind MW, cost of under estimation and the cost of X
N X
NW X
NS
over estimation with ‘NW’ number of wind generators. Vi Vj Yij cosðdi  di  hij Þ þ PWind;i þ PS;i
i¼1 i¼1 i¼1
X
NW
CW ¼ ½CðPWind; i Þ þ CWU; i þ CWO ;i  $=h ð19Þ ¼ PD þ PL
i¼1 ð27Þ
where ‘CðPWind;i Þ’ is the cost of ith wind generator to
generate ‘PWind’. MW can be expressed as
2. Reactive power balance
CðPWind;i Þ ¼ KW  PWind;i ð20Þ
The total reactive power generation must be equal to the
Because of the involvement of RES, it is most uncertain sum of system reactive demand and reactive transmission
to guess their generation. To incorporate the uncertain losses.
effect, the mathematical modeling includes the cost of Mathematically,
overestimation and the cost of under estimation.
X
nl
The cost of under estimation (CWU, I) and overestima- Vi Vj Yij sinðdi  di  hij Þ þ Qsh ¼ QD þ QL ð28Þ
tion (CWO, I) can be expressed as i¼1
 
CWU;i ¼ KRW;i  PW Avail;i  PWind;i ð21Þ where ‘Vi’ and ‘Vj’ are the bus voltages in p.u at buses i and
  j; ‘Yij’ is the YBus matrix in p.u; ‘di’ and ‘dj’ are the bus
CWO;i ¼ KPW;i  PWind;i  PW Avail;i ð22Þ
voltage angles in radians at bus i and j; ‘hij ’ is the admit-
where ‘PW Avail;i ’ is actually available wind generation; tance angle in radians; ‘PD’ is the active power demand in
‘KW’ = direct cost coefficient. ‘KRW,i’ = reserve cost p.u. ‘PL’ is the active power loss in p.u; ‘Qsh’ is the shunt
coefficient for under estimation;‘KPW,I’ = penalty cost reactive power in p.u; ‘QD’ is the reactive power demand in
coefficients for overestimation of ith wind generator; p.u; ‘QL’ is the reactive power loss in p.u;
The cost of solar generator (CS) includes cost to gen-
b. System inequality Constraints
erate ‘PS’ MW, cost of under estimation and the cost of
1. Bus voltages must be within specified limits.
overestimation with ‘NS’ number of solar plants.
X
NS
CS ¼ ½CðPS; i Þ þ CSU; i þ CSO ;i  $=h ð23Þ jVimin j  jVi j  jVimax j; i ¼ 1; 2; 3; . . .NB ð29Þ
i¼1

123
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79 67

2. The active power and reactive power generation must faster, convergence particle weight factor Pwf is intro-
be within the specified limits. duced. After testing on a different standard functions, the
range of Pwf varies between 0 and 2 (Venkata Rao and
Pmin  Pi  Pmax ; i ¼ 1; 2; 3; . . .NG ð30Þ Patel 2013). The modified air velocity equation can be
i i
expressed as
 
1  
Utþ1 ¼ ð1  aw ÞUt  gxt þ RT  1 xopt  Pwf  xt
Qmin
i  Qi  Qmax
i ; i ¼ 1; 2; 3; . . .NG ð31Þ r
cuotherdim
t
3. The wind and solar power generation should be within þ ð36Þ
the feasible region r
The new position of air particle will be updated as
PWind ¼ 0; m\mCut in xtþ1 ¼ xt þ Uiþ1 ð37Þ
¼ 0:5  q  A  m3 ; mCut in \m\mCut out ð32Þ where ‘a’ is constant; ‘Ut’ is the present velocity of air
¼ Pr ; m [ mCut out parcels; ‘g’ is the gravitational constant; ‘xt’ is the current
position of air particle; ‘RT’ is the coefficient; r is a con-
stant; ‘xopt’ is the so far obtained optimal value; ‘cuotherdim
t ’
PS ¼ 0; GHI \ GHImin is the Contribution of Coriolis force
¼ PR  g  GHI  AS  Sz; GHI [ GHImin
ð33Þ 8 Solution Methodology
Where ‘m’ = wind velocity; ‘mCut in ’ = cut-in speed
The following steps are to be implemented for giving OPF
of a wind turbine; ‘mCutout ’ = cut-out speed of a wind
problem on a given system in different scenarios.
turbine; ‘q’ = density of air; ‘AW’ = wind turbine
blade area of cross section. ‘Pr’ = rated power; Step 1: Consider the windiest areas in the AP 14 bus
‘GHI’ = global horizontal irradiance in KW/m2; system.
‘GHImin’ = minimum value of GHI; ‘PR’ = perfor- Step 2: Forecast the wind speed in identified areas for
mance ratio of solar panel. ‘g’ = efficiency of solar short-term time period (20 h) and long-term time period
panel; ‘AS’ = area of solar panel; ‘Sz’ = size of solar (15 years) with the help of proposed P-ELM approach.
plant. Step 3: With the forecasted wind speed values, calculate
4. The power flow in a transmission line is limited by its the wind power available in these areas using Eq. (12).
thermal capability. Step 4: Similarly, consider the four areas which are
having more sunny hours using the CI method.
Sl  Smax ; l ¼ 1; 2; 3; . . .nl ð34Þ Step 5: Forecast the GHI available in the considered
l
areas for short-term time period and long-term time
5. The tap setting of a regulating transformer should be period using proposed P-ELM approach.
within its maximum and minimum limits. Step 6: Calculate the solar power that can be produced
using Eq. (15) in the considered locations.
Timin  Ti  Timax ; i ¼ 1; 2; 3; . . .t ð35Þ Step 7: Solve the OPF problem using IWDA without
RES and obtain the solution, which is a base case
solution and extend this procedure for short term period.
Step 8: For a long-term period, vary the load demand in
7 Improved Wind-Driven Algorithm (IWDA) steps of 4.5% per year and solve the OPF up to a
maximum number of years.
Wind-driven optimization is established depending on Step 9: Integrate the wind power into the system and
motion of air particles in the atmosphere because of uneven solve the OPF problem for the considered short-term and
pressure distribution (Bayraktar et al. 2013). It is governed long-term time periods in four areas for all seasons.
by Newton’s second law. The moment of air pockets is Step 10: Integrate the solar power into the system and
directed by the frictional forces raised by neighboring solve the OPF problem for the considered short-term and
particles, gravitational force, which draws air parcels to long-term time periods in four areas for all seasons.
center of search space, the force of pressure gradient and Step 11: Integrate both the wind and solar powers into
forces caused by Coriolis. To make the algorithm much the system and solve the OPF problem for the considered

123
 Author's personal copy
68 Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79

short-term and long-term time periods in four areas for In the first step, the wind- and solar-rich locations are
all seasons. identified as described in Sect. 5. Using the proposed
Step 12: Analyze the variation of voltages at different P-ELM approach wind power and solar powers are fore-
buses, power flows, total fuel cost and transmission casted in the identified areas. The obtained statistical
losses in all cases. results are validated with the existing literature, methods
reported in Zhang et al. (2017) and are tabulated in
Tables 1 and 2, respectively. From this forecasting of wind
with the proposed method, MAE and RMSE are decreased
9 Results and Analysis
by 3% and 0.83% when compared to (Zhang et al. 2017).
The solar insolation is also forecasted using proposed
In this section, the entire analysis is performed on a real-
P-ELM approach, and the obtained statistical results are
time Andhra Pradesh 14-bus system.
validated with the existing literature method reported in
Yousif et al. (2017), Huang et al. (2016) and Yona et al.
(2013) which are tabulated in Table 2. From this, the
Table 1 Comparison of statistical metrics with existing and proposed proposed method MAE is decreased by 46.2%, 94.2% and
method for wind speed forecasting 95.71% when compared to Yousif et al. (2017) and 11.7%
Algorithms MAE RMSE when compared to Yona et al. (2013). At the same time, the
RMSE of the proposed method is decreased by 94.4%
ANFIS (Zhang et al. 2017) 0.15 0.17 compared to Huang et al. (2016).
SARIMA (Zhang et al. 2017) 0.14 0.18 From the above two tables, it is clear that the proposed
Zhang et al. (2017) 0.05 0.06 P-ELM algorithm yields better forecasting results with less
Persistent 0.0498 0.0496 error when compared to existing methods. The simulation
ELM 0.0491 0.0524 has been carried out with the aid of MATLAB 2013a. The
Proposed method 0.0485 0.0490 main contribution of hybrid P-ELM approach is not only
accurate forecasting over other statistical approaches, but
also the computation is done over a shorter time horizon.
For the considered real-time AP system, the simulation
Table 2 Comparison of statistical metrics with existing and proposed time taken by P-ELM is 0.4155 s, which excludes training
method for solar insolation forecasting
and determining of the output weights with a PC of 2 GB
Algorithms MAE RMSE RAM Intel i5 processor, while it is 0.7077 and 0.4282 s for
the persistence and ELM, respectively. Therefore, the
SOF (Yousif et al. 2017) 0.361 –
computation time for P-ELM is considerably less com-
MML (Yousif et al. 2017) 3.369
pared to others.
PSVM (Yousif et al. 2017) 4.537
SVM (Huang et al. 2016) – 5.6
9.1 OPF Results—Variation of Power Loss
FL-RNN (Yona et al. 2013) 0.22 –
and Cost in Different Cases
Persistent 0.2145 0.3247
ELM 0.2018 0.3102
The OPF problem is solved with the integration of RES in
Proposed P-ELM 0.1942 0.3096
the following 12 cases and the supporting numerical and
Active power loss (MW)

4.7 800

4.6 780
Cost ($/hr)

760
4.5
740
4.4
720
4.3 700
4.2 680
1 2 3 4 1 2 3 4
1-Case (i) 2-Case(iv) 3-Case (vii) 4-Case(x) 1-Case(i) 2-Case(iv) 3-Case(vii) 4-Case(x)

Fig. 7 Active power loss and fuel cost for the cases (1), (4), (7) and (10)

123
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79 69

Short term Short term

Hourly total active power

7 970
920
6.5

Cost ($/hr)
870
6
820
loss (MW)

5.5
770
5 720
4.5 670
4 620
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Time (Hours) Time (Hour)

Case (ii) Case (v) Case (viii) Case (xi) Case (ii) Case (v) Case (viii) Case (xi)

Long term Long term

Total active power loss

5.4 840
5.3
820
5.2

Cost ($/hr)
800
5.1
(MW)

5 780
4.9 760
4.8
740
4.7
4.6 720
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Duration (Years) Duration (Years)

Case (iii) Case (vi) Case (ix) Case (xii)

Case (iii) Case (vi) Case (ix) Case (xii)

Fig. 8 OPF solution: short-term and long-term period for cases (3), (6), (9) and (12)

79 20500000
78
Active power loss (MW)

77 20000000

76
Cost ($/hr)

19500000
75
74
19000000
73
72
18500000
71
70 18000000
1 2 3 4
1 2 3 4
1-Case (i) 2-Case(iv) 3-Case (vii) 1-Case(i) 2-Case(iv) 3-Case(vii) 4-Case(x)
4-Case(x)

Fig. 9 Active power loss and fuel cost for the cases (1), (4), (7) and (10)

graphical analysis is presented for the considered two test short term period is considered to be 20 h) for further
systems. analysis.
3. Without incorporation of RES for a long-term time
1. Without incorporation of RES-base case analysis.
period of 15 years. Here, the load variation is
2. Without incorporation of RES for a short-time
considered 4.5% per year. (As per the last 10-year
period, where electrical load variation has been
data collected from Vidyut Soudha, Hyderabad).
taken for 20 h (for the considered system, the load
4. By integrating wind generation into the system-base
flow solution is converged up to 20 h. Hence, the
case with wind power.

123
 Author's personal copy
70 Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79

Short term
25000000 Short term

Hourly total acve power oss in

24000000
400
23000000

Cost ($/hr)
22000000
300
21000000
20000000

MW
200
19000000
18000000
100
17000000
16000000
0

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Time (Hour)
Case (ii) Case (v) Case (viii) Case (xi) Case(ii) Case(v) Case(viii) Case(xi)

Long term Long term

22000000 92
21800000

Total acve power loss (MW)

21600000 90
21400000
88
21200000
Cost ($/hr)

21000000 86
20800000
84
20600000
20400000 82
20200000
20000000
80
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
78
Duration (Years) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Case (iii) Case (vi) Case (ix) Case (xii)

Case (iii) Case (vi) Case (ix) Case (xii)

Fig. 10 OPF solution: short-term and long-term period for cases (3), (4), (9) and (12)

Base case Short term

16 LOAD FLOW OPF

Active power loss (MW)

Hourly total active power loss

14
12 20

10
15
8
10
(MW)

6
4
5
2
0 0
1 2 1 3 5 7 9 11 13 15 17 19
1- Load flow 2-OPF Time (Hours)

Long term

LOAD FLOW OPF

18
Total active power loss (MW)

16
14
12
10
8
6
4
2
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Duration (Years)

Fig. 11 Loss comparison for OPF and load flow without integrating RES

123
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79 71

Short term
Base case
778 OPF LOAD FLOW
776
774 950
772
Cost ($/hr)

900
770

Cost ($/hr)
768 850
766 800
764
750
762
760 700
758 650
1 2 1 3 5 7 9 11 13 15 17 19

1- Load flow 2- OPF Time (Hours)

Long term

LOAD FLOW OPF

2000

1800

1600

1400
Cost ($/hr)

1200

1000

800

600

400

200

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Duration (years)

Fig. 12 Cost comparison for OPF and load flow without integrating RES

5. By integrating wind generation for a short-term time 11. By integrating wind and solar generation for a short-
period of 20 h. term time period of 20 h.
6. By integrating wind generation for a long-term time 12. By integrating wind and solar generation for a long-
period of 15 years. term time period of 15 years.
7. By integrating solar generation into the system-base
case with solar.
9.1.1 AP 14-Bus Syetm
8. By integrating solar generation for a short-term time
period of 20 h.
In this section the OPF problem is executed for AP 14-bus
9. By integrating solar generation for a long-term time
system and the results are obtained for the aforementioned
period of 15 years.
12 cases. The findings of the simulations are demonstrated
10. By integrating wind and solar generation into the
in graphical representation.
system-base case with wind and solar powers.

123
 Author's personal copy
72 Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79

Base Case Short term

14

Hourly total active Power loss (MW)

LOAD FLOW OPF

Active power loss (MW)

12
18
10 16
14
8
12
6 10
8
4
6
2 4
2
0
1 2 0
1 2 3 4 5 6 7 8 9 1 01 11 21 31 41 51 61 71 81 92 0
1- load flow 2- OPF
Time (Hours)

Long term

LOAD FLOW OPF

Total active power loss (MW)

16
14
12
10
8
6
4
2
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Duration (Years)

Fig. 13 Loss comparison for OPF and load flow—with wind integration

Base case Short term

756 LOAD FLOW OPF

754
950
Cost ($/hr)

752
Cost ($/hr)

900
750
850
748
800
746 750
744 700
742 650
1 2 1 3 5 7 9 11 13 15 17 19
1- LOAD FLOW 2- OPF Time (Hours)

Long term
LOAD FLOW OPF
806

804

802
Cost ($/hr)

800

798

796

794

792
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Duration (Years)

Fig. 14 Cost comparison for OPF and load flow with wind integration

123
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79 73

Base case Short term

14

Hourly total active power loss

Active power loss (MW)
LOAD FLOW OPF
12
18
10 16
14
8
12

(MW)
6 10
8
4 6
2 4
2
0 0
1 2 1 3 5 7 9 11 13 15 17 19

1- Load Flow 2- OPF Time (Hours)

Long term

LOAD FLOW OPF

16
Total active power loss (MW)

14

12

10

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Duration (Years)

Fig. 15 Loss comparison for OPF and load flow with solar integration

From Fig. 7, it can be observed that the active power loss integration of combined wind and solar, the losses and cost
is less in the base case, but it is more for the case wind both are much less compared to remaining cases.
integrated only. The solar only and wind, solar combination
yields to less cost in the order. And it is also observed that the 9.1.2 IND 124-Bus Test System
operating cost is higher in the base case and less for wind and
solar combination combination. In this section, the OPF problem is executed for IND-124
Figure 8 shows the OPF solution for short-term as well bus system and the results are obtained for the aforemen-
as long-term period of 15 years in different cases. The tioned 12 cases. The findings of the simulations are
graphs are drawn for loss and as well as for cost parame- demonstrated in graphical representation.
ters. It is understood that, during the short term period the Figure 9 shows the active power loss and fuel cost for a
cost is high for a case (2), which is not integrating RES base case for IND 124-bus system. It is observed that the
case and low in case (12), which is both wind and solar loss is minimum in base case and high in the case of a wind
integration cases. The loss also results in more value in integrated system. This is due to the fact that the objective
case (3), where there is no RES integration and less value in OPF is considered as cost minimization rather than loss.
in case (12), which is an integration of both wind and solar And it is also observed that the cost is less in case of both
combined. Therefore, it can be concluded that with the wind and solar integrated to grid.

123
 Author's personal copy
74 Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79

Base case Short term

756 LOAD FLOW OPF

754
950
752
Cost ($/hr)

900

Cost ($/hr)
750
850
748
800
746
750
744
700
742
650
1 2
1 3 5 7 9 11 13 15 17 19
1- LOAD FLOW 2- OPF
Time (Hours)

Long term

LOAD FLOW OPF

806

804

802
Cost ($/hr)

800

798

796

794

792
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Duration (Years)

Fig. 16 Cost comparison for OPF and load flow—with solar integration

Figure 10 shows the fuel cost and loss variations in the 9.2 Comparison of Load Flow and OPF Solution
time periods of long-term and short-term time horizons. In
both cases, the cost is high and losses are less in base case In this section, the comparison of results of load flow and
and both wind and solar integrated scenario, respectively. OPF problem is presented for one system (AP-14 bus) to
This is due to the fact that the loads which are nearer to the highlight the importance of OPF over load flow.
RES are met by them instead of drawing from central
a. Without integrating RES:
sources.
To show the significance of OPF, the above analysis is The OPF problem is solved in the cases of without and
also carried out with and without RES integration and the with RES integration. The obtained results are compared
results are compared with that of load flow solution in the with the results obtained from load flow solution in terms
subsequent sections. of transmission losses and cost. From Figs. 11 and 12, it is
observed that the losses and cost both become less for OPF
in the base case, short term and long term also compared to

123
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79 75

Base case Short term

Active power loss (MW)

14

Hourly total active power

LOAD FLOW OPF
12
10 20
8
15

loss (MW)
6
4 10
2 5
0
0
1 2 1 3 5 7 9 11 13 15 17 19

1 - Load Flow 2 - OPF Time (Hours)

Long term

LOAD FLOW OPF

16
Total active power loss (MW)

14

12

10

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Duration (Years)

Fig. 17 Loss comparison for OPF and load flow with wind and solar integration

that of load flow. The reason is in OPF problem; the gen- c. With integrating solar power:
erators are rescheduling their generations to minimize the
In this section, the OPF problem is solved by integrating
cost as well as loss.
solar power. From Figs. 15 and 16, it is observed that the
b. With integrating Wind power: losses and cost both become less for OPF in the base case,
short term and long term as well compared to that of load
In this section, the OPF problem is solved by integrating
flow. The reason behind this is the solar power meets the
wind power. From Figs. 13 and 14, it is observed that, the
loads which are nearer to its installed location. Therefore,
losses and cost both become less for OPF in the base case,
the burden of supplying the load from the conventional
short term and long term as well compared to that of load
generators reduces, which results in less cost and also less
flow. The reason is the wind generation meets the loads
loss.
which are near its location. Therefore, the burden of sup-
plying the load from the conventional generators reduces, d. With the integration of wind and solar power:
which results in less cost and less loss also.
In this section, the OPF problem is solved by integrating
both wind and solar powers. From Figs. 17 and 18, it is
observed that the losses and cost both become less for OPF

123
 Author's personal copy
76 Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79

Short term
Base case
730 LOAD FLOW OPF
728
726
Cost ($/hr)

724
820
722

Cost ($/hr)
720 770
718
716 720
714
670
712
710 620
1 2 1 3 5 7 9 11 13 15 17 19
Time (Hours)
1- Load Flow 2- OPF

Long term

LOAD FLOW OPF

780

778

776

774
Cost ($/hr)

772

770

768

766

764

762

760
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Duration (Years)

Fig. 18 Cost comparison for OPF and load flow with wind and solar integration

in the base case, short term and long term as well compared of RES penetrations has been analyzed along with sup-
to that of load flow. The generated wind and solar power porting numerical and graphical results. From the results, it
meet the loads which are nearer to their installed location. has been identified that due to rescheduling of generations
Therefore, the necessity of supplying the load from the in economic aspect, the total generation fuel cost has been
conventional generator decreases, which results in less cost reduced and also the total power losses have also been
and less loss as well. decreased when compared to load flow with the RES
integration. Finally, with the integration of wind and solar
combination, rewarding results can be achieved for a
10 Conclusion considered time period.

A new methodology and a novel IWDA have been pre-

sented in this work to solve OPF problem. With this, the Appendix
complexity in solving OPF problem with the integration of
RES in the grid has been minimized. The problem of See Tables 3, 4, 5, 6 and 7.
violating inequality constraints has been solved, and effect

123
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79 77

Table 3 Bus data of AP 14-bus system

Bus Area Type Voltage Theta Pgen Qgen P load Q load Qmin Qmax Qsh
no. (p.u) (degrees) (MW) (MVAR) (MW) (MVAR) (MVAR) (MVAR) (MVAR)

1 Slack 1 1.06 0 0 0 0 0 10 250 0

2 Krishna 2 1.045 0 40 0 19.7 12.7 -40 50 0
3 Nellore 2 1.01 0 0 0 90.2 18 0 40 0
4 West Godavari 3 1 0 0 0 45.8 - 3.9 0 0 0
5 Guntur 3 1 0 0 0 6.6 1.6 0 0 0
6 Kadapa 2 1.07 0 0 0 10.2 7.5 - 66 24 0
7 East Godavari 3 1 0 0 0 8.2 2.8 0 0 0
8 VSKP 2 1.09 0 0 0 11.8 1.2 -6 24 0
9 Prakasam 3 1 0 0 0 28.5 14.6 0 0 19
10 Chittoor 3 1 0 0 0 7 4.8 0 0 0
11 Kurnool 3 1 0 0 0 2.5 1.8 0 0 0
12 Anantapur 3 1 0 0 0 5.1 1.6 0 0 0
13 Srikakulam 3 1 0 0 0 11.5 5.8 0 0 0
14 Vijayanagaram 3 1 0 0 0 11.9 5 0 0 0

Table 4 Line data of AP 14-bus system

Line no. From area To area Resistance (p.u) Reactance (p.u) Line charging (B/2) (p.u) Tap (p.u) Line limit (MVA)

1 Slack Krishna 0.01938 0.05917 0.0264 1 150

2 Slack Guntur 0.05403 0.22304 0.0246 1 85
3 Krishna Nellore 0.04699 0.19797 0.0219 1 85
4 Krishna West Godavari 0.05811 0.17632 0.0187 1 85
5 Krishna Guntur 0.05695 0.17388 0.017 1 85
6 Nellore West Godavari 0.06701 0.17103 0.0173 1 85
7 West Godavari Guntur 0.01335 0.04211 0.0064 1 150
8 West Godavari East Godavari 0 0.20912 0 0.98 30
9 West Godavari Prakasam 0 0.55618 0 0.97 32
10 Guntur Kadapa 0 0.25202 0 0.93 45
11 Kadapa Kurnool 0.09498 0.1989 0 1 14
12 Kadapa Anantapur 0.12291 0.25581 0 1 32
13 Kadapa Srikakulam 0.06615 0.13027 0 1 22
14 East Godavari VSKP 0 0.17615 0 1 32
15 East Godavari Prakasam 0 0.11001 0 1 29
16 Prakasam Chittoor 0.03181 0.0845 0 1 32
17 Prakasam Vijayanagaram 0.12711 0.27038 0 1 18
18 Chittoor Kurnool 0.08205 0.19207 0 1 12
19 Anantapur Srikakulam 0.22092 0.19988 0 1 12
20 Srikakulam Vijayanagaram 0.17093 0.34802 0 1 12

123
 Author's personal copy
78 Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79

Table 5 Generator cost coefficient and their limits data of AP 14-bus system
S. no. Area A ($/MW2h) b ($/MWh) c ($/h) Pgmin (MW) Pgmax (MW) Qmin (MVAR) Qmax (MVAR)

1 Slack 0.00375 2 0 10 250 10 250

2 Krishna 0.0175 1.75 0 10 200 - 40 50
3 Nellore 0.0625 1 0 5 60 0 40
4 Kadapa 0.00834 3.25 0 5 50 -6 24
5 Visakhapatnam 0.025 3 0 5 60 -6 24

Table 6 Technical specifications of wind turbine Huang C, Bensoussan A, Edesess M, Tsui KL (2016) Improvement in
artificial neural network-based estimation of grid connected
Suzlon 2.1 MW, S97 wind turbine model photovoltaic power output. Renew Energy 97:838–848
Kathiravan R, Kumudini Devi RP (2017) Optimal power flow model
Parameter Value incorporating wind, solar, and bundled solar-thermal power in
Rated power 2.1 MW the restructured indian power system. Int J Green Energy
Cut-in wind speed 3.5 m/s 14(11):934–950
Khaligh A, Onar OC (2017) Energy harvesting: solar, wind, and
Rated wind speed 11 m/s ocean energy conversion systems. CRC Press, Boca Raton
Cut-out wind speed 20 m/s Khan B, Singh P (2017) Optimal power flow techniques under
Rotor diameter 97 m characterization of conventional and renewable energy sources:
a comprehensive analysis. J Eng Hindawi 2017:1–16
Swept area 7386 m2
Khazal A, Kalantar M (2015) Optimal generation dispatch incorpo-
Hub height 90 m rating wind power and responsive loads: a chance-constrained
framework. J Renew Sustain Energy 7(2):1–22
Lahari NV, Vasantha Kumar Shetty H (2015) integration of grid
connected PMG wind energy and solar energy systems using
different control strategies. In: International conference on
Table 7 Technical specifications of solar power panel
power and advanced control engineering (ICPACE)
Solar panel specification Lotfy ME, Senjyu T, Farahat MA-F, Abdel-Gawad AF, Lei L, Datta
M (2018) Hybrid genetic algorithm fuzzy-based control schemes
Maximum power 250 Wp for small power system with high-penetration wind farms. Appl
Operating voltage 0–40 V Sci 8(3):1–20
Patel MR (2005) Wind and solar power systems: design, analysis, and
Type of solar panel Monocrystalline, polycrystalline
operation. CRC Press, New York
Dimensions 2000 mm 9 1000 mm Peng C, Zou J, Zhang Z, Han L, Liu M (2017) An ultra-short-term
pre-plan power curve based smoothing control approach for grid-
connected wind-solar-battery hybrid power system. IFAC Pap
Online 50(1):7711–7716
Sekar K, Duraisamy V (2015) Efficient energy management system
References for integrated renewable power generation system. J Sci Ind Res
74:325–329
Ackermann T (2012) Wind power in power systems. Willey, New Shilaja C, Ravi K (2016) Optimal power flow considering intermittent
York wind power using particle swarm optimization. Int J Renew
Alasan Martin PC, Nikitovic L, Mujovic S (2019) CONOPT solver Energy Res 6(2):504–509
embedded in GAMS for optimal power flow. J Renew Sustain Sunil Kumar C, Subramanyan PS, Amarnath J (2012) Available
Energy 046301(11):1–16 transfer capability calculations with contingencies for Andhra
Bayraktar Z, Komurcu M, Bossard JA, Werner DH (2013) The wind Pradesh state grid. Int J Sci Eng Res 3(6):1–7
driven optimization technique and its application in electromag- Surender Reddy S (2017) Optimal power flow with renewable energy
netics. IEEE Trans Antennas Propag 61(5):2745–2757 resources including storage. Electric Eng 99(2):685–695
Biswas PP, Suganthan PN, Amaratunga Gehan AJ (2017) Optimal Syed MS, Suresh C, Sivanagaraju S (2017) Short term solar insolation
power flow solutions incorporating stochastic wind and solar prediction: p-ELM approach. Int J Parallel Emerg Distrib Syst
power. Energy Convers Manag 148(2017):1194–1207 33(6):663–674
Cardenas R, Perez MA, Clare JC (2017) Control and grid integration Teeparthi K, Vinod Kumar DM (2017) Multi-objective hybrid PSO–
of MW-range wind and solar energy conversion systems. IEEE APO algorithm based security constrained optimal power flow
Trans Ind Electron 64(11):8786–8789 with wind and thermal generators. Eng Sci Technol Int J
Chamanbaz M, Dabbeney F, Lagoaz C (2017) AC optimal power flow 20(2):411–426
in the presence of renewable sources and uncertain loads. IEEE Teeparthi K, Vinod Kumar DM (2018) Security-constrained optimal
Trans Power Syst 1–22 power flow with wind and thermal power generators using fuzzy
Duncan Glover J, Overbye T, Sarma MS (2016) Power system adaptive artificial physics optimization algorithm. Neural Com-
analysis and design, 6th edn. Cengage Learning put Appl 29(3):855–871

123
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Electrical Engineering (2021) 45:61–79 79

Venkata Rao R, Patel V (2013) An improved teaching-learning-based network for long-term ahead PV power output correction. IEEE
optimization algorithm for solving unconstrained optimization Trans Sustain Energy. 4(2):527–533
problems. Sci Iran D 20(3):710–720 Yousif JH, Kazem HA, Boland J (2017) Predictive models for
Wood AJ, Wollenberg BF (2013) Power generation, operation and photovoltaic electricity production in hot weather conditions.
control. Wiley, New York Energies 10(7):971
Yona A, Senjyu T, Funabashi T, Kim CH (2013) Determination Zhang J, Wei Y, Tan Z-f, Ke W, Tian W (2017) A hybrid method for
method of ınsolation prediction with fuzzy and applying neural short-term wind speed forecasting. Sustainability 9(4):1–10

123