Received May 30, 2020, accepted June 4, 2020, date of publication June 9, 2020, date of current version June

24, 2020.
Digital Object Identifier 10.1109/ACCESS.2020.3000988

Photovoltaic Array Reconfiguration System for

Maximizing the Harvested Power Using
Population-Based Algorithms
THANIKANTI SUDHAKAR BABU 1,2 , (Member, IEEE), DALIA YOUSRI 3,

AND KARTHIK BALASUBRAMANIAN 4 , (Member, IEEE)

1 Department of Electrical Power Engineering, Institute of Power Engineering, Universiti Tenaga National (UNITEN), Kajang 43000, Malaysia
2 Department of Electrical and Electronics Engineering, G. Pullaiah College of Engineering and Technology, Kurnool 518002, India
3 ElectricalEngineering Department, Faculty of Engineering, Fayoum University, Fayoum 63514, Egypt
4 Offshore Technology Development Private Ltd., Keppel Offshore and Marine Ltd., Singapore 629353

Corresponding author: Dalia Yousri ([email protected])

ABSTRACT Massive infiltration of photovoltaic (PV) systems into electric supply networks creates numer-
ous challenges in the present era, as the PV systems become an alternative to non-renewable energy resources.
Partial shading, nevertheless, is an essential problem which affects the productivity and life of PV plants.
PV reconfiguration is known as a powerful technique to resolve this effect. It is achieved by rearranging
the PV modules according to their temperature and levels of shade. Therefore, in this paper, we have
utilized three simple population-based optimization algorithms that are known as the flow regime algorithm
(FRA), the social mimic optimization algorithm (SMO), and the Rao optimization algorithm to dynamically
restructure the PV array. The effectiveness of the proposed algorithms is evaluated using several metrics such
as fill factor, mismatch losses, percentage of power loss, and percentage of power enhancement. Besides,
the results obtained are compared with a regular total-cross-tied (TCT) connection and recently published
techniques such as the competence square (CS) and genetic algorithm (GA). Furthermore, to demonstrate
the suitability of proposed approaches in real-time implementation, real-time irradiation data of a particular
location are considered and fed into the proposed algorithms for effective shade dispersion. After successful
shade dispersion, the total energy generated using the three proposed algorithms is calculated and compared
with the TCT reconfigured system for one year. The presented energy calculations and revenue generation
confirm that the power produced by the proposed FRA technique is 13% higher than that generated by the
TCT configuration. Furthermore, the presented PV characteristics show a reduced number of multiple peaks
in the system. Thus, the proposed FRA technique can be endorsed as a technique that is superior to other
existing methods.

INDEX TERMS PV reconfiguration, partial shading, mitigation techniques, flow regime algorithm, social
mimic optimization, Rao optimization algorithm.

NOMENCLATURE GOA grasshopper optimization

ACRONYMS HC honey comb
BL bridge link MHHO modified harris hawks optimizer
CS competence square NR newton-raphson
FRA flow regime algorithm PLPB piece-wise linear parallel branches
GA genetic algorithm PSO particle swarm based algorithm
GMPP global maximum power point PV photovoltaic
SDS shade dispersion scheme
SMO Social mimic optimization algorithm
The associate editor coordinating the review of this manuscript and SP series parallel
approving it for publication was N. Prabaharan . TCT total cross tied

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
109608 VOLUME 8, 2020
T. S. Babu et al.: PV Array Reconfiguration System for Maximizing the Harvested Power

VARIABLES Vt thermal voltage

g∗nt and lnt
∗ best and the worst solutions at itera- Varray total PV array voltage
tion nt Vmx PV module voltage at row x
ISC short circuit current Vmx PV module voltage at row x
Zi(nt ) solution vector of particle i at the cur- VOC open circuit voltage
rent iteration nt We and Wparray weight factors of Pe and Parray
β scaling factor Zi(nt ) solution vector of the ith particle at
γ constant its value is 1.5 iteration nt
a1 ideality factor Zint and Zjnt solution of ith candidate and the jth
Arraypower total array power candidate
D1 Diode Zi , Zj , and Zk particles with the numbers i, j and k
fitness(i) fitness value of the ith element Pe summation of the error difference
G actual irradiation
g∗nt global best solution obtained until the
nth
t iteration I. INTRODUCTION
g∗nt global best solution that is obtained In the present world, renewable energy resources have gained
until the nth
t iteration extreme solicitude because of different key reasons, including
GMPPchallenger GMPP attained by the TCT, Rao, the consumption of non-renewable energy sources, atmo-
or SMO-configured system sphere concerns, furthermore, the penchant to have a greenery
GMPPendorsed global power produced by the recom- and amicable atmosphere [1]. The photovoltaic (PV) inno-
mended configuration vation and development is viewed as a monetarily serious
GMPPSTC GMPP at STC inexhaustible force innovation in the short term. Due to the
GMPPPSC GMPP at PSC development of technology in the PV field, the installation
ID1 current flowing through the diode cost of PV will be reduced day by day. Therefore, the instal-
Iirr PV current source lation of PV plants around the globe by the year 2018 is
Imp and Vmp maximum produced values of current 512 GW, it shows the commitment and utilizing the solar PV
and voltage under the PSC source effectively. However, the growth of PV is increasing,
Io1 diode D1 leakage current the enhancing efficiency of the PV system is yet a challeng-
Ip1 current across the shunt resistance ing task due to its nonlinear characteristics and completely
IRx PV array total current depends on atmospheric conditions [2]. The other important
Isc short-circuit current factor which affects the degradation of PV power is partial
Ix , Vx current and voltage of the PV array for shading, this occurs mainly due to building shadows, dust
the x th row and bird droppings, manufacturing defects, and nonuniform
IC fully irradiated condition ageing of PV modules [3].
k Boltzmann constant = 1.3805 − Reconnecting the PV modules in different configurations
10−23 J /K can help in reducing the effect of partial shading via distribut-
ki current coefficient factor ing the shadow in the array surface. With this inspiration, the
Le0 vy number generated by the Levy distri- various basic connections schemes like series-parallel (SP),
bution function total cross-tied (TCT), honeycomb (HC) and bridge link (BL)
Ns number of cells in series were introduced [4]. Later, Gautam et al., in [5] used proba-
nt present iteration number bility theory to assess the reliability of large scale PV array
Parray PV array output power without using three different connections such as SP, TCT and BL.
bypassing By performing extensive computation analysis of array size
PMax(IC) generated maximum power during the of 720 × 20, authors confirmed that by introducing cross
fully irradiated condition ties between PV modules, the operational lifetime of TCT
Ppeak maximum value of the power connected system has increased by 30% than regular con-
q electron charge = 1.6 − 10−19 C nections. Similarly, authors in [6] evaluated the performance
Rp1 shunt resistance of the basic interconnection schemes with the size of 6 × 6
Rs1 series resistance PV array under six various irradiation patterns, it is found
rand random number withdrawn from the that TCT is the one, which enhances the power under any
Gaussian distribution shade conditions than other basic connection schemes. Fur-
rand1 and rand2 random numbers withdrawn from the thermore, Belhachat et al. in [7] also worked on basic con-
interval (0,1) nection schemes over an array of 6 × 4 and proved as TCT is
T cell temperature in Kelvin the most suitable connection under partial shade conditions.
Tmax max. no. of iterations Besides, in [8], authors scrutinized the electrical performance
u, v random values between 0 to 1 of PV arrays of SP, TCT, HC and BL configurations with

VOLUME 8, 2020 109609

 T. S. Babu et al.: PV Array Reconfiguration System for Maximizing the Harvested Power

the help of Newton–Raphson (NR) and the piece-wise linear Dynamic electrical PV array connection scheme reduces
parallel branches (PLPB) methods successfully. The TCT the partial shading effect, notifying the status of the PV
method shows superior performance in two considered meth- system at every moment [20]. However, the published patents
ods. However, the authors have evaluated the technique under on dynamic reconfiguration system boost up further research
one shade condition, which is insufficient to reach inferences in this field to enhance the power generation and improve the
on the best connection scheme. On the complete outcomes efficiency of the PV system under any shade conditions [21].
acquired by the above literature, it demonstrates that the A power comparison based reconfiguration for SP and TCT
prevalence of the TCT connection exhibits good performance connected system is detailed in [15]. Authors in [22], recon-
in maximum cases of partial shading scenarios. On the other figured the TCT connected modules such that, equalizing the
side, it is observed that the output current generated by TCT short circuit current of each row, with this approach authors
connection is limited if the maximum amount of PV modules reduced the number of switches in the proposed system.
in a row were shaded [9], due to this power generated by PV In addition, in this method, TCT connected 4 × 4 PV array is
system get reduces greatly. Thereby it opens the challenge to split into a fixed and dynamic part, as a fixed portion, authors
find other suitable techniques. considered 3 × 3 PV array which will not show betterment
After carrying out rigorous research, authors found a new in the system performance. Further, the authors in [23], the
technique by changing the position of PV module’s or altering proposed novel method named optimized string dynamic PV
connections between PV modules to disperse shade over array to reduce the effect of partial shading. In this method,
PV array equally which enhances power generation, it is PV strings with the same irradiation level are connected to an
known as PV reconfiguration. El-Dein et al.in [9] introduced individual converter. However, this method is expensive since
mathematical formulation based on mixed-integer quadratic it consumes a high number of converter circuits.
programming to disperse the shade of a TCT connected To overcome the limitations of dynamic reconfiguration
system. Authors tested the proposed method by considering methods, as perfect choice, evolutionary algorithms have
the full, half re-configurable PV array and TCT connected been introduced to reconfigure the PV modules based on irra-
system. Fixed reconfiguration scheme for 3 × 3 PV array diation levels. To the best of authors’ familiarity, evolutionary
is proposed in [10]. The proposed concept shows efficient algorithms have been successfully implemented in various
performance than SP, TCT, and BL configurations. Adap- fields of applications [24]–[26]. With this motivation authors
tive reconfiguration technique is introduced by Nugen et. al. in [19], [27] proposed genetic (GA) and particle swarm
in [11], this follows an easier control technique. However, based algorithm (PSO) to configure the switching matrix
this method consumes more number of switches and sensors based on irradiation levels. In the above mentioned works
which increases the system cost. As of late, a shade dispersion authors aimed to maintain the row current difference between
scheme (SDS) in light of a fixed electrical topology has been each row as equal using regular weighted objective function.
proposed in [12] to increase the power generation of PV Later, Fathy in [28], proposed novel grasshopper optimiza-
plant. This method mainly works on renumbering the PV tion (GOA) based algorithm with same objective function.
modules based on the output of the algorithm, accordingly, The performance of the proposed method is evaluated with
the electrical connections of PV modules will be altered. 9 × 9 PV array under various shade patterns. The introduced
In addition to the above-discussed methods, the static optimization based reconfiguration methods work effectively
reconfiguration methods were introduced which effectively in individual, moreover, they exhibits limitations such as GA
solves the problem of partial shading and improves the power method consumes more time as it involves large computation
generation capability of PV plant. In static reconfiguration, steps for convergence, also it has chances of stuck at local
the position of PV modules will be changed physically. The peak. Further, PSO has nature of premature convergence
various static reconfiguration methods recently introduced which will confuse the system behaviour. Tuning parameters
by authors are given in [13]–[18]. In static reconfiguration used in GA, PSO and GOA are the challenging task, improper
methods, the dispersion of shade mainly takes place based on selection of tunable parameters shows huge impact on power
puzzle outcome. Moreover, this is an onetime arrangement production. Due to inconsistency behaviour of GA and PSO
of PV structure, according to the considered shade pattern. operating under dynamic change in irradiation conditions will
If the new shade occurs the system has to rebuild again and not give favourable solution. By observing the limitations of
alter the position of PV modules manually. So this type of above methods, recently Dalia et.al., in [29] developed a new
arrangements is not reliable as the irradiation will not be technique named modified harris hawks optimizer (MHHO)
constant in a practical scenario. In addition, these methods with scheme of prey catching by hawks. In this work, authors
consume lengthy cables while changing the position of PV executed extensive analysis by considering various array
modules physically which increases additional line losses and structures such as 9 × 9, 6 × 4 and 6 × 20. The obtained
also require skilled labour to perform the job [19]. However, results from this method shows better performance than other
static methods show superior performance over traditional methods discussed earlier. Even though authors used existing
hitherto known interconnection topologies, the static methods objective function, the performance of the PV system has
are not favoured in real-time due to above said limitations. been enhanced with unique features of MHHO.

109610 VOLUME 8, 2020

T. S. Babu et al.: PV Array Reconfiguration System for Maximizing the Harvested Power

Applying the meta-heuristic for PV array reconfiguration

approach is a newly trend and proposing an efficient algo-
rithm for switching matrix design is a challenge task. There-
fore, with intention of providing the complete solution to
the effect of partial shading, in this work authors developed
three recent successful algorithms that are SMO, FRA, Rao
for the application of PV array reconfiguration at first time.
The three proposed algorithms are selected because they
have the features of the simplicity in the implementation
and requiring low number of tuned permeates. To evaluate
these optimization techniques, a weighted objective function
has been considered and tested under various shade patterns
FIGURE 1. Electrical circuit of single diode PV model.
including with real time irradiation data available over period
of ten years. The key findings of this article are:
• The authors implement three population-based solar PV in it. In similar, the performance of PV system is also depends
reconfiguration algorithms. on its effective modelling of PV cell. Modelling of solar PV
• The flow regime algorithm (FRA), SMO algorithm, and cell is a challenging task due to its non-linear behaviour [30].
Rao optimization algorithms are implemented for PV Thereby, various authors modelled different type of PV mod-
reconfiguration of 9 × 9 PV array subjected for three ules, such as one diode, two diode and three diode models.
shade patterns including long with broad shading, short Among these, one (single) diode model attains high attention
with narrow shading and long with narrow shading. due to less parameters involved and easy to model [31].
• Efficiency metrics are computed and compared with the Thereby, most of the researchers preferred construct their PV
TCT and other established methods in literature such as modules with single diode PV model. In this work authors
competence square approach (CS) and particle swarm used single diode model to design the PV module and PV
optimizer (PSO) in order to check the superiority of array size of 9 × 9. The equivalent electrical circuit of consid-
proposed population-based algorithms. ered PV model is shown in Fig. 1, which constitute a current
• Real-time average irradiation data for 10 years during source Iirr connected anti-parallel to the to the diode D1 ,
March are taken into account and fed into optimization series and shunt resistances are Rs1 and Rp1 .
algorithms as one shade pattern from 7 AM to 8 PM per The current generated by PV cell can be estimated by
hour. applying Kirchhoff current law to the equivalent circuit. The
• The power produced and revenue generated are esti- total current delivered by single-diode PV model can be
mated for 1 year and compared with the TCT perfor- given as:
mance.
• PV characteristics are plotted between the TCT connec- IPV = Iirr − ID1 − Ip1 (1)
tion and proposed techniques to show the reduction of
where Iirr is a PV current source, ID1 is current flowing
multiple peaks in the PV plant
through the diode, and Ip1 is current across the shunt resis-
The analyses across the studied shade patterns reveal on the tance. By substituting ID1 and Ip1 , the equation of the current
superiority of FRA technique in providing the highest power can be written as follows:
values and offering a regular dispersion for the shade patterns  
VD1
 
V + IRs1

over the surface of the array. IPV = Iirr − Io1 exp −1 − (2)
a1 Vt Rp1
The remainder of the article structure is formulated as
follows. The II-A section defines the PV module modeling. where Vt is the thermal voltage and can be depicted as NsqKT ,
In Section II-B, the representation of the TCT connected where k is the Boltzmann constant = 1.3805Ã − 10−23 J /K ,
PV array and its current-voltage calculations are provided. T is the cell temperature in Kelvin, q is the electron charge
In Section III, explanations of the proposed algorithms and = 1.6Ã − 10−19 C, Ns is the number of cells in series, a1 is
implementation steps are described. In the IV portion, the the ideality factor, and Io1 is the diode D1 leakage current.
formulation of the objective function of the proposed tech- The power generated by a PV system greatly depends on
niques are given. The results and discussions are shown in environmental conditions. Therefore, the current generated
the V section. The findings and conclusion of the carried-out by a PV source can be mathematically represented as given
work are presented in Section VI. in Eq. 3.
 
G
II. SYSTEM DESCRIPTION Iirr = [Isc + ki (T − T0 )] (3)
A. MODELING OF PV MODULE G0
The performance and efficiency of any system is mainly where Isc represents the short-circuit current at standard test
depends on effective design of its basic components involved condition (STC), that is, G0 = 1000W /m2 and T0 = 25◦ C.

VOLUME 8, 2020 109611

 T. S. Babu et al.: PV Array Reconfiguration System for Maximizing the Harvested Power

algorithms for the application of PV array reconfiguration to

enhance the power generation are discussed in the following
sections.

III. PROPOSED POPULATION-BASED ALGORITHMS

The population-based algorithms have proven their superi-
ority in solving nonlinear multi-modal optimization prob-
lems compared with individual optimization algorithms.
Consequently, in the current work, several techniques have
been implemented for PV reconfiguration to appraise their
responses and recommend the most appropriate technique.
The no-free-lunch theory mentioned that no algorithm is con-
sidered a super technique for all optimization problems [32],
which motivated the authors to test numerous algorithms.
In addition, the chosen algorithms are simple in their exe-
cution with different inspiration fields. The implemented
algorithms are the Rao optimization algorithm [33], SMO
algorithm [34], and FRA [35]. The details of the proposed
techniques are discussed next.

A. FLOW REGIME ALGORITHM

FRA was recently developed based on the basic concepts of
flowing fluids in basic fluid mechanics [35]. Flowing fluids
FIGURE 2. TCT connected 9 × 9 PV array structure.
can take on two profiles: laminar and turbulent. The authors
G and T are considered actual irradiation and temperature in [35] considered these flowing categories in programming
values, respectively, and ki indicates the current coefficient their algorithm, in which laminar flow describes a local
factor. search (exploitation stage) while turbulent flow is responsible
for a global search (exploration phase). To achieve a balance
B. TCT-CONNECTED PV ARRAY between the local and the global phases of a search, the
TCT connection scheme can be framed by connecting cross authors suggested using a switching factor between the two
ties to the series parallel connected system. It is the most stages, named the search type factor (STF) [35]. The STF can
widely used system to reduce the effect of partial shading and be computed as follows:
|| g∗nt − Zi ||


reducing the losses over other basic connections, same can Tmax
6
be observed from the presented literature. Therefore, in this STFi = 3.2 × 10 × ×
(6)
nt || Zj − Zk ||
article authors developed a 9×9 TCT connected PV array for
the evaluation of proposed reconfiguration techniques. The where g∗nt is the global best solution, which is obtained until
designed 9 rows and 9 columns TCT connected PV array is the ntht iteration, nt is the present iteration number, and Tmax
shown in Fig. 2. Each module in the PV array is indicated is the max. no. of iterations. The symbols of Zi , Zj , and Zk
by x, y, which shows the rows and columns numbers of PV denote particles with the numbers i, j and k, respectively.
modules. If the STF has values ≥ 3.2 × 105 , the algorithm follows the
The total current and voltage of a TCT connected system diversification stage, whereas if STF is less than 3.2 × 105 ,
can be calculated as given in Eq. 4 and Eq. 5, respectively. the intensification stage is the dominant one.
9
For the global search (diversification), the authors used the
IRx =
X
Ixy , x = 1, 2 . . . 8, 9 (4) following equation:
y=1 0.37
Zi(nt +1) = Zi(nt ) + βLe0 vy × g∗nt − Zi(nt ) × √


5
(7)
9
X STF
Varray = Vmx (5) where Zi(nt ) is the solution vector of particle i at the current
x=1 iteration nt . β is a scaling factor is taken as 30%. Le0 vy is a
where IRx is the PV array total current produced at row x, number generated by the Levy distribution function based on
Varray is the total voltage that appears across terminals of the the following equation:
PV array, and Vmx is the PV module voltage at row x. 1
γ

As TCT connected system fails to scatter the shade over πγ
u × σ 0(1 + γ ) × sin( )
entire PV array, in this work authors proposed new opti-
0
Le vy = 0.01 × , σ= 2
( γ −1
 
1
mization techniques which effectively disperse the shade of |v| γ 0( 1+γ
2 )×γ ×2
2 ))

a TCT connected system. The implementation of proposed (8)

109612 VOLUME 8, 2020

T. S. Babu et al.: PV Array Reconfiguration System for Maximizing the Harvested Power

FIGURE 4. Flowchart of SMO.

FIGURE 3. Flowchart of FRA.

where u, v are random values between 0 to 1, and γ is constant

its value is 1.5.
The local search in the algorithm is performed using the
following equation:
4.96
Zi(nt +1) = Zi(nt ) + βrand × g∗nt − Zi(nt ) × √


(9)
STF
where rand is the random number withdrawn from the Gaus-
sian distribution. The flowchart of FRA is in Fig. 3.

B. SMO ALGORITHM
Human facial expressions and body signals inspired the
authors in [34] to implement the SMO algorithm. SMO
programs the performance of an individual who trying to
assimilate into a society of influential people by emulating FIGURE 5. Flowchart of Rao.
their style. The authors used the difference function between
the optimal solution for an individual and that obtained at the
last iteration to update the solution vector using the following achieved during the optimization process and the random
equations: function (rand) to interconnect the solutions. The main con-
trol equation of the Rao technique is as follows:
g∗nt − objectivefunction Zint



fDiffnt+1 = (10a) Zint+1 = Zint +rand1 g∗nt −|lnt
∗



objectivefunction Zint

| + rand2 |Zint | − Zjnt (11)
Zint+1 = Zint + fDiffnt+1 × Zint (10b) where rand1 and rand2 are random numbers withdrawn from
the interval (0,1); g∗nt and lnt∗ are the best and the worst
where fDiffnt+1 is the difference function that is computed
solutions at iteration nt, respectively; and Zint and Zjnt are the
based on the first formula in Eq.10a when the obtained value
solution of ith candidate and the jth candidate that are selected
is equal to 0, the fDiffnt+1 value is withdrawn from the random
randomly, respectively. The flowchart in Fig. 5 shows the
distribution between (0,1). g∗nt is the global best solution that
execution steps of the Rao technique.
is obtained until the nth t iteration, Zi(nt ) is the solution vector
of the ith particle at iteration nt , and objective function is the IV. OBJECTIVE FUNCTION DEFINITION
implemented fitness function that should be minimized or Defining the objective function is one of the main require-
maximized. To clarify the steps of the SMO algorithm, its ments for the optimization algorithm to start the process.
flowchart is shown in Fig. 4. For the PV array reconfiguration problem, to achieve the
maximum power of the reconfigured PV array, the objective
C. RAO OPTIMIZATION ALGORITHM
function is framed as follows:
The Rao optimization algorithm was proposed following
a new trend in population size optimization algorithms do Maximize (fitness(i))
 
not depend on inspiration from nature. The Rao optimiza- We

= Arraypower + + Wparray × Parray (12)
tion algorithm depends on the best and the worst solutions Pe
VOLUME 8, 2020 109613
 T. S. Babu et al.: PV Array Reconfiguration System for Maximizing the Harvested Power

Pseudocode of Reconfiguration Based on Optimization Algo- robust and reliable algorithm that achieves the optimal PV
rithm reconfiguration scheme. The algorithms have been imple-
Inputs: The irradiance levels, Vx , and Ix . Population size mented for 300 iteration numbers and 50 population size.
N and number of iterations T The simulations and analysis were executed using laptop with
Outputs: The optimal recombination of the switching specifications of 16 GB RAM, Core i7, 2.5 GHz of speed
matrix processor and version of MATLAB 2018.
Initialize the swarm matrix with dimensions of N , and d. 1) The introduced population-based algorithms’ perfor-
d represents the unknown variables. mance was examined over three profiles of partial shad-
while (termination criteria is not met) do ing, with long, broad distribution for shadow over the
Calculate the fitness values of particles via Eq. 12. considered PV array, as well short and long narrow
Set g∗ as the best arrangement for the PV modules in shade patterns. The shade patterns were implemented
the considered array. on a 9 × 9 PV array interconnected in a TCT arrange-
for (each particle (Zi )) do Update the swarm matrix ment. The generated interconnected systems based on
based on the algorithm control equations (FRA; Eqs. 7 & 9. the introduced techniques were compared with recently
SMO; Eq. 10b. Rao; Eq. 11). proved approaches such as CS and GA, as well as the
end for basic scheme of the TCT configuration. The compari-
end while son focused on the generated power, PV characteristics
Return Optimal reconfiguration g∗ and the attained mismatch power loss, the fill factor, the
percentage of power loss, and the percentage of power
enhancement.
where fitness(i) represents the fitness value of the ith element 2) The proposed Rao optimization algorithm, SMO algo-
in the present population. Arraypower is the total array power, rithm, and FRA were tested experimentally for a PV
and it can be defined as given in Eq. 13: plant with a rated capacity of 4.585 kW constructed
with a 9 × 9 PV array. The examination was carried
9
X out with 12 shade patterns throughout the day from
Arraypower = Ix × Vx (13)
8 AM to 7 PM with a step 1 hour. By applying the
x=1
proposed algorithms to the developed PV plant, the
where Ix and Vx are the current and voltage of the PV array energy generated by each method was calculated and
for the x th row, respectively. compared with the traditional TCT-connected system.
Pe is the summation of the error difference between the In addition, the amount of revenue generated and the
peak power of the obtained PV characteristics and the other amount of energy saved using the proposed techniques
power values for each row. It can be calculated as shown in were presented. To perform the energy calculations, the
Eq. 14: cost per unit of energy was considered to be INR 15,
9 as provided by the authors in [27].
X
Ppeak − Ix × Vx ,
 
Pe = (14)
x=0 A. SIMULATION AND DISCUSSION
where Ppeak is the maximum value of the power and Ix ×Vx is The Rao optimization algorithm, SMO algorithm, and FRA
constitutes the other local values of the x th row power. Using are applied to provide the optimal recombination for the
the difference between the global obtained power value and switching matrix over three levels of shading.
the local ones helps in attaining a reliable dispersion for the
shading levels, which subsequently appears as only one peak 1) LONG WITH WIDE SHADING (PATTERN 1)
in the PV characteristics. The shadow in pattern 1 was distributed broadly with sev-
Parray is the PV array output power without bypassing. eral irradiance levels (900, 600, 500, 400, and 200 W /m2 )
We and Wparray are weight factors for Pe and Parray , respec- over several modules of the considered array. The rows and
tively, and they are taken as 10. columns with the numbers 7, 8, and 9 were fully shad-
The Rao optimization, SMO, and FRA optimization tech- ing with levels of 600, 500, and 400W /m2 , respectively,
niques are implemented using the described fitness func- whereas the other modules were subjected to 900W /m2 .
tion to identify the optimal reconfiguration for the switching The considered shade pattern of the TCT-connected scheme
matrix based on the following steps as shown in following is shown in Fig. 6(a). Similarly, the dispersed shade pat-
pseudo code of algorithm. terns after successful implementation of the reconfiguration
techniques of CS, GA, Rao, SMO, and FRA are presented
V. RESULTS AND ANALYSIS in Fig. 6(b) to 6(f) respectively. The Rao, SMO and FRA
Several analysis and comparisons were performed over the consumed about 1 second and half until reaching for the
proposed algorithms (Rao optimization algorithm, SMO depicted configuration, whereas the GA consumed longer
algorithm, and FRA) to evaluate and demonstrate the most time of 4 second.

109614 VOLUME 8, 2020

T. S. Babu et al.: PV Array Reconfiguration System for Maximizing the Harvested Power

TABLE 1. Analysis of TCT, CS, GA, Rao, SMO, and FRA for long with wide shading condition.

After performing the reconfiguration using the proposed  for the 6th row
– The rowcurrent  can be 
calculated
 as
900 600 500
methods, the shade patterns for pattern 1 were obtained as IR6 = 3 1000 IM + 1 1000 IM + 1 1000 IM +
presented in Fig. 6. According to the reconfigured structure,
   
400 200
3 1000 IM + 1 1000 IM = 5.2IM .
the row currents in each technique can be calculated using
Eq. 4.  for the 7th row
– The rowcurrent  can be calculated as
900 400
IR7 = 6 1000 IM + 3 1000 IM = 6.6IM .
• The calculation of row currents for the TCT scheme
 for the 8th row
– The rowcurrent  can be 
calculated
 as
considered in Fig. 6(a) can be given as follows: 900 500 400
IR8 = 4 1000 IM + 1 1000 IM + 1 1000 IM +
– For the first six rows, the currents have the same
 
200
3 1000 IM = 5.1IM .
value, which are
 computed asshown:
– The rowcurrent for the 9th row
 can be 
calculated
 as
 
900 400
IR1 to IR6 = 6 1000 IM + 3 1000 IM = 6.6IM . 900 600 400
IR9 = 4 1000 IM + 1 1000 IM + 3 1000 IM +
– Row currents for rows 7, 8, and 9 can be given as  
200
follows:     1 1000 IM = 5.6IM .
600 500
IR7 = IR8 = IR9 = 2 1000 IM + 2 1000 IM +
    The row currents for the proposed Rao scheme presented in
400 200 these equations by following the similar procedure currents
2 1000 IM + 3 1000 IM = 3.6IM .
of reaming schemes can also be evaluated. With the aid of
• The calculation of row currents for the proposed Rao calculated row currents, the power generated by each row
scheme as in Fig. 6(d) can be given as follows: can be evaluated and is exhibited in Table 1. From the pre-
 for the 1st row
– The rowcurrent  can be 
calculated
 as sented Table, it can be understood that the proposed FRA
900
IR1 = 4 1000 IM + 2 1000 500 400
IM + 1 1000 IM + attains high power level, with a value of 49.5VM IM . It also
  contains fewer variations in row currents. Therefore, power
200
2 1000 IM = 5.4IM . generated by the proposed FRA can be enhanced. To prove
– The row current for the 2nd row can be calculated the amount of power enhancement, the current-voltage (I-V)
as       and P-V curve of the dispersed shade patterns are presented
900 600 500 in Fig. 7. The presented P-V curves show reduced multiple
IR2 = 4 1000 IM + 1 1000 IM + 1 1000 IM +

400
 peaks because of an equal shade dispersion over the PV array.
3 1000 IM = 5.9IM .
In addition, it generates power of 2731 W, which is higher
 for the 3rd row
– The rowcurrent  can be calculated
 as than that of other schemes. However, competitive methods
900 600 400
IR3 = 4 1000 IM + 1 1000 IM + 4 1000 IM = such as CS, GA, and TCT generate 2716 W, 2726 W, and 2128
5.8IM . W of power, respectively. Even though the power difference
 for the 4th row
– The rowcurrent  can be calculated
 as between the proposed FRA and the competitive methods is
900 600 400
IR4 = 4 1000 IM + 1 1000 IM + 2 1000 IM + less, it shows high variation in the case of highly rated PV
 
200
2 1000 IM = 5.4IM . plants.
– The rowcurrent for the 5th row
 can be 
calculated
 as
900
IR5 = 3 1000 IM + 1 1000 600 500
IM + 1 1000 IM + 2) SHORT WITH NARROW SHADING (PATTERN 2)
  With shade pattern 2, irradiance levels of 900W /m2 ,
400
4 1000 IM = 5.4IM . 600W /m2 , and 400W /m2 are incident on the surface of the

VOLUME 8, 2020 109615

 T. S. Babu et al.: PV Array Reconfiguration System for Maximizing the Harvested Power

FIGURE 6. Short wide shadow pattern for pattern 1: (a) TCT, (b) CS [14], (c) GA [19], (d) proposed Rao, (e) developed SMO, and
(f) developed FRA.

TABLE 2. Analysis of TCT, CS, GA, Rao, SMO, and FRA for short with narrow shading condition.

considered array. The first six rows were fully illuminated by modules, and 600W /m2 was received by the other four
900W /m2 while the seventh and eighth rows were partially modules in the row. The considered shade pattern of the
shaded by 600W /m2 , and 400W /m2 . For the last row in TCT-connected scheme and dispersed shade after implemen-
the array, the irradiance of 900W /m2 covered the first five tation of the reconfiguration schemes are presented in Fig. 8.

109616 VOLUME 8, 2020

T. S. Babu et al.: PV Array Reconfiguration System for Maximizing the Harvested Power

FIGURE 7. PV characteristics obtained by TCT, CS, GA, Rao, SMO, and FRA for pattern 1: (a) P-V curves and (b) I-V curves.

FIGURE 8. Short wide shadow pattern for pattern 2: (a) TCT, (b) CS [14], (c) GA [19], (d) proposed Rao, (e) proposed SMO, and
(f) developed FRA.

The Rao, SMO and FRA achieved those patterns in range of As discussed in Section V-A1, row currents and power lev-
1.1 − 1.2 second whilst the GA has a longer execution time els for each scheme were calculated and are given in Table 2.
of 3.8 second. From the table, it can be noticed that the proposed methods

VOLUME 8, 2020 109617

 T. S. Babu et al.: PV Array Reconfiguration System for Maximizing the Harvested Power

FIGURE 9. PV characteristics obtained by TCT, CS, GA, Rao, SMO, and FRA for pattern 2: (a) P-V curves and (b) I-V curves.

of Rao optimization, SMO, and FRA generate power levels are presented in Table 3. The computations in Table 3 show
of 63VM IM , 63.9VM IM , and 64.8VM IM , respectively, with that, the proposed FRA scheme generates only two variations
fewer variations in row current. To show the superiority of the in power levels and attains high values of the power level
proposed methods in practice, the I-V and P-V curves of the 63.9VM IM . The reduced power variations reflect the number
proposed schemes in comparison with the TCT, GA, and CS of multiple peaks in P-V curves. The other methods consist
methods are displayed in Fig. 9. Fig. 9 exhibits the effective of a greater number of power variations, which indicates a
shade dispersion using the proposed population-based recon- high number of peaks; therefore, power produced by the PV
figuration schemes. In addition, the proposed FRA technique plant can be reduced. To indicate the multiple peaks and
generates 3731 W, whereas the TCT, GA, and CS tech- generated maximum power, the I-V and P-V characteristics
niques generate 3501 W, 3710 W, and 3713 W, respectively, of pattern 3 for the considered methods are shown in Fig. 11.
which is less power compared with the proposed techniques. From Fig. 11, it can be understood that, the proposed FRA
In addition, the zoomed-in version of I-V and P-V curves technique generates a higher amount of power than other
are presented, which clarifies the closeness of the curves to techniques and is close to that of the GMPP.
the global maximum power point (GMPP). Therefore, the It can be concluded that among the three proposed
introduced FRA technique can track GMPP effectively in schemes, the FRA technique exhibits superior performance
shorter execution time in comparison with GA and other in three types of shade conditions. It also generates a high
competitors. amount of power close to the GMPP value, with fewer peaks
in the PV characteristics, furthermore, the execution time of
3) LONG WITH NARROW SHADING (PATTERN 3) FRA is the shorter one in comparable with GA and the other
For pattern 3, the shadow of four irradiance levels is dis- counterparts. This indicates the consistency and effectiveness
tributed over three areas of the considered array. The last of the proposed FRA technique.
two modules of the third, fourth, and fifth rows are subjected
to 700W /m2 , whereas irradiance levels of 400W /m2 and 4) EFFICIENCY METRICS
300W /m2 covered the last three modules of the sixth and In this section, several measures, namely mismatch power
seventh rows and the eighth and ninth rows, respectively. loss, fill factor, percentage of power loss, and percentage
The surface of the other modules received 900W /m2 . The of power enhancement are computed over the studied shade
considered shade pattern of the TCT-connected scheme and patterns to determine and enhance the most efficient and
dispersed shade after implementation of the reconfiguration suitable algorithm: the Rao optimization algorithm, SMO
schemes are presented in Fig. 10. The execution time of the algorithm, or FRA. The same metrics are computed for GA,
proposed algorithms until providing those patterns are 1.3, CS, and TCT interconnected arrangements based on the listed
1.35, 1.1 and 4.1 seconds for Rao, SMO, FRA and GA, results in Sections V-A1, V-A2, and V-A3 for comparison.
respectively. The mathematical formulas of the used factors can be written
Furthermore, to show the consistency of the proposed tech- as follows:
niques, another shade pattern named long with narrow shade Mismatch power loss
was considered. Similar to the results in Section V-A1, the
row currents and obtained power levels were calculated and = PMaxIC − PGMPPPSC (15)

109618 VOLUME 8, 2020

T. S. Babu et al.: PV Array Reconfiguration System for Maximizing the Harvested Power

TABLE 3. Analysis of TCT, CS, GA, Rao, SMO, and FRA for long with narrow shading condition.

FIGURE 10. Short wide shadow pattern for pattern 3: (a) TCT, (b) CS [14], (c) GA [19], (d) proposed Rao, (e) proposed SMO, and
(f) developed FRA.

Fill factor Percentage of power loss

(Vmp Imp )PSC GMPPSTC − GMPPPSC

= (16) = (17)
VOC ISC GMPPSTC
VOLUME 8, 2020 109619
 T. S. Babu et al.: PV Array Reconfiguration System for Maximizing the Harvested Power

FIGURE 11. PV characteristics obtained by TCT, CS, GA, Rao, SMO, and FRA for pattern 3: (a) P-V curves and (b) I-V curves.

FIGURE 12. Mismatch power loss for the studied shade patterns: (a) pattern 1, (b) pattern 2, and (c) pattern 3.

FIGURE 13. Fill factor for the studied shade patterns: (a) pattern 1, (b) pattern 2, and (c) pattern 3.

Percentage of power enrichment of current and voltage under the PSC, respectively. ISC
GMPPchallenger − GMPPendorsed and VOC represent the short circuit current and open cir-
= (18) cuit voltage, respectively. GMPP at STC and PSC are
GMPPchallenger
represented by GMPPSTC and GMPPPSC , respectively.
Here, IC is a fully irradiated condition. PMax(IC) and GMPPchallenger emphasizes the GMPP attained by the
PGMPP(PSC) are the generated maximum power values dur- TCT, Rao, or SMO-configured system, while GMPPendorsed
ing the fully irradiated condition and at PSC, respec- is the global power produced by the recommended
tively. Imp and Vmp are the maximum produced values configuration.

109620 VOLUME 8, 2020

T. S. Babu et al.: PV Array Reconfiguration System for Maximizing the Harvested Power

FIGURE 14. Percentage of power loss for the studied shade patterns: (a) pattern 1, (b) pattern 2, and (c) pattern 3.

To showcase the superiority of the proposed metaheuristic-

based reconfiguration methods, the three considered patterns
were analyzed using the previously mentioned formulas.
To present an effective comparison, the results obtained with
proposed algorithms were compared with previously pub-
lished techniques, such as the TCT, CS, and GA reconfigu-
ration. As performance indicators of the system, mismatch
losses have high priority, because they give the difference
of power generation between fully irradiated conditions and
PSCs. This represents the amount of power loss reduced
using the proposed techniques. To represent this, total power
generation during fully irradiated conditions (real global
power), power generated during shade conditions and mis-
match power loss for three considered patterns are presented
in Fig. 12. Fig. 12 shows that the presented techniques (TCT,
CS, GA, Rao optimization, and SMO) reach high values in FIGURE 15. Percentage of power enhancement between FRA
the bar diagrams, which indicates high amounts of power loss arrangement and other approaches.
in the system. However, the FRA technique attains smaller
values under the three shade patterns. Therefore, it can be Finally, after performing rigorous analysis, the percent-
confirmed that the proposed FRA technique is a good solution age of power enhancement for the three shade patterns
for the problem of partial shading effects. with other interconnection schemes was evaluated and is
Other notable performance indicators, such as the fill presented in Fig. 15. Among the presented techniques, the
factor, percentage of power loss, and percentage of power combined FRA-reconfiguration technique shows significant
enrichment, were calculated for the three shade patterns and power enhancement compared with other techniques in all
are presented in Fig. 13, Fig. 14, and Fig. 15, respectively. shade patterns. It achieves 28.333% in pattern 1, 6.588% in
The fill factor is an important parameter that can be used pattern 2, and 6.998% in pattern 3.
to assess the excellence of the PV plant in poorly irradiated From the presented analysis and discussions, it can be
conditions. A high value of the fill factor shows a greater concluded that from the three proposed metaheuristic tech-
effectiveness of the PV systems; this can be observed in niques, FRA can be endorsed as the superior technique. The
Fig. 13 for the three shade patterns. The proposed FRA FRA-reconfiguration technique gives a solid solution for the
technique attains a high fill factor value, which indicates the effect of partial shading even under fast and dynamic change
superiority of the technique compared with others. Similarly, in irradiation conditions.
the percentage of power loss for the three shade patterns is
exhibited in Fig. 14. From Fig. 14(a), 14(b), and 14(c), it can B. COMPREHENSIVE EVALUATION
be observed that the TCT-connected reconfiguration system To estimate the consistency and efficiency of the proposed
generates a high percentage of power losses. Furthermore, the reconfiguration techniques, a test was carried out by consid-
differences in percentage of power loss among the proposed ering the real-time irradiation that occurred in Coimbatore,
techniques is lower, which will lead to higher amounts of Tamil Nadu, India, with a latitude and longitude of 11.0168◦
power among highly rated PV plants. Therefore, the pro- N, and 76.9558◦ E. The average irradiation levels for a
posed FRA technique succeeded in greatly reducing power duration of 10 years were considered, using a period from
losses. 7 AM to 8 PM. To test the proposed system, the real-time

VOLUME 8, 2020 109621

 T. S. Babu et al.: PV Array Reconfiguration System for Maximizing the Harvested Power

FIGURE 16. Real-time irradiation pattern and obtained P-V curves: (a) irradiation levels, (b) shade patterns, (c) P-V curves accomplished using TCT
without reconfiguration, and (d) P-V curves after successful implementation of FRA for 9 × 9 PV array.

irradiation levels were fed into the developed 9 × 9 PV array. produces energy 13% higher than that of the TCT configura-
The 9 × 9 connected PV array was designed by considering tion. Furthermore, SMO and Rao optimization reconfigura-
the PV modules of SM55, and the total rated capacity of the tion techniques produce energy 11% and 9% higher than that
system was 4.455 kW. The variation in irradiation over a day of TCT-connected system, respectively. These results prove
for the considered duration is presented in Fig. 16(a). that the proposed FRA technique greatly enhances power
The test was carried out hourly for a specified period. generation. Furthermore, the TCT-connected reconfiguration
The irradiation for each hour for a TCT-connected system is generates an income of USD 1139.73 per year, whereas FRA
shown in Fig. 16(d). produces an income of US 1284.90.
For the same shade patterns, the proposed algorithms were Furthermore, partial shading effect can be clearly observed
implemented to scatter the shade over the PV array. After per- from the PV characteristics of any PV system. The P-V
forming the dispersion every hour, the energy was calculated curves for the 12 considered shade patterns are presented
for each shade pattern using four algorithms, the TCT config- in Fig. 16(c). From Fig. 16(c), it is observed that numer-
uration, FRA, the SMO algorithm, and the Rao optimization ous peaks exist. The presence of multiple peaks reduces
algorithm, to indicate the power enhancement using the pro- the power generation drastically. As a solution to the par-
posed techniques. Table 4 shows the energy calculated from tial shading effect, the proposed FRA technique was imple-
real-time data using the proposed metaheuristic-based recon- mented; with this technique, the apparent shade over the
figuration techniques. From the table, it can be understood PV array was effectively dispersed over the entire PV array.
that the proposed metaheuristic-based FRA reconfiguration As a result of effective dispersion, the multiple peaks over

109622 VOLUME 8, 2020

T. S. Babu et al.: PV Array Reconfiguration System for Maximizing the Harvested Power

TABLE 4. Analysis of energy production among reconfiguration presented. The presented results confirm that among the three
techniques.
proposed population-based algorithms, the FRA technique
achieves superior performance and high power enhancement
with the fewest power losses. It offers effective optimization
and higher flexibility for reconfiguration approaches where
it converged for the optimal structure in 1.5 second whilst
the GA consumed about 4 second. Moreover, the introduced
techniques defeats the occurrence of multiple peaks in-case
of high shading conditions, which proved higher power gen-
eration than other methods. Hence, the FRA technique can
be endorsed as an effective method to minimize the partial
shading effect in large-scale PV plants.

REFERENCES
[1] D. Yousri, T. S. Babu, D. Allam, V. K. Ramachandaramurthy, and
M. B. Etiba, ‘‘A novel chaotic flower pollination algorithm for global max-
imum power point tracking for photovoltaic system under partial shading
conditions,’’ IEEE Access, vol. 7, pp. 121432–121445, 2019.
[2] M. Azharuddin Shamshuddin, T. S. Babu, T. Dragicevic, M. Miyatake,
and N. Rajasekar, ‘‘Priority-based energy management technique for inte-
gration of solar PV, battery, and fuel cell systems in an autonomous DC
microgrid,’’ Electr. Power Compon. Syst., vol. 45, no. 17, pp. 1881–1891,
Oct. 2017.
[3] D. Yousri, T. S. Babu, D. Allam, V. K. Ramachandaramurthy, E. Beshr, and
M. B. Eteiba, ‘‘Fractional chaos maps with flower pollination algorithm
for partial shading mitigation of photovoltaic systems,’’ Energies, vol. 12,
P-V characteristics were reduced. It is well known that the no. 18, p. 3548, Sep. 2019.
reduction of multiple peaks increases the number of units [4] R. Ramaprabha and B. L. Mathur, ‘‘A comprehensive review and analysis
generated. To highlight the number of reduced peaks before of solar photovoltaic array configurations under partial shaded conditions,’’
Int. J. Photoenergy, vol. 2012, pp. 1–16, Mar. 2012.
and after implementation of FRA, a column in Table 4 shows [5] N. K. Gautam and N. D. Kaushika, ‘‘Reliability evaluation of solar photo-
the number of peaks that appeared in the TCT configura- voltaic arrays,’’ Sol. Energy, vol. 72, no. 2, pp. 129–141, Feb. 2002.
tion and the proposed FRA. Furthermore, the obtained P-V [6] O. Bingöl and B. Özkaya, ‘‘Analysis and comparison of different PV array
configurations under partial shading conditions,’’ Sol. Energy, vol. 160,
characteristics using FRA are presented in Fig. 16(d). The pp. 336–343, Jan. 2018.
Fig. 16(d) shows the number of multiple peaks was reduced in [7] F. Belhachat and C. Larbes, ‘‘Modeling, analysis and comparison of solar
the system. This indicates the amount of power enhancement photovoltaic array configurations under partial shading conditions,’’ Sol.
Energy, vol. 120, pp. 399–418, Oct. 2015.
using the proposed FRA technique. In addition, the anal- [8] Y.-J. Wang and P.-C. Hsu, ‘‘An investigation on partial shading of PV
ysis performed for energy savings and revenue generation modules with different connection configurations of PV cells,’’ Energy,
acknowledges the superiority of the proposed FRA technique. vol. 36, no. 5, pp. 3069–3078, May 2011.
[9] M. Z. Shams El-Dein, M. Kazerani, and M. M. A. Salama, ‘‘Optimal
photovoltaic array reconfiguration to reduce partial shading losses,’’ IEEE
VI. CONCLUSION Trans. Sustain. Energy, vol. 4, no. 1, pp. 145–153, Jan. 2013.
Partial shading is one major phenomenon that deteriorates the [10] P. Srinivasa Rao, G. Saravana Ilango, and C. Nagamani, ‘‘Maximum
power from PV arrays using a fixed configuration under different shading
generation of the maximum power from PV plants. To mini- conditions,’’ IEEE J. Photovolt., vol. 4, no. 2, pp. 679–686, Mar. 2014.
mize the partial shading effect, various approaches have been [11] D. Nguyen and B. Lehman, ‘‘An adaptive solar photovoltaic array using
introduced in the literature, such as using bypass diodes, dis- model-based reconfiguration algorithm,’’ IEEE Trans. Ind. Electron.,
vol. 55, no. 7, pp. 2644–2654, Jul. 2008.
tributed MPPT, and static reconfiguration techniques. How- [12] P. R. Satpathy, R. Sharma, and S. Jena, ‘‘A shade dispersion interconnection
ever, these techniques fail to implement in large-scale PV scheme for partially shaded modules in a solar PV array network,’’ Energy,
plants because of their limitations. Therefore, this article pro- vol. 139, pp. 350–365, Nov. 2017.
[13] S. G. Krishna and T. Moger, ‘‘Optimal SuDoKu reconfiguration technique
poses new population-based dynamic reconfirmation tech- for total-cross-tied PV array to increase power output under non-uniform
niques to provide a permanent solution for the problem of irradiance,’’ IEEE Trans. Energy Convers., vol. 34, no. 4, pp. 1973–1984,
partial shading. The authors successfully propose three opti- Dec. 2019.
[14] B. Dhanalakshmi and N. Rajasekar, ‘‘A novel competence square based PV
mization algorithms, namely, FRA, SMO, and Rao optimiza-
array reconfiguration technique for solar PV maximum power extraction,’’
tion algorithms for providing the optimal reconfigured PV Energy Convers. Manage., vol. 174, pp. 897–912, Oct. 2018.
structure for 9 × 9 PV array subjected for three different [15] M. Akrami and K. Pourhossein, ‘‘A novel reconfiguration procedure to
shade patterns. Different performance measures are calcu- extract maximum power from partially-shaded photovoltaic arrays,’’ Sol.
Energy, vol. 173, pp. 110–119, Oct. 2018.
lated such as fill factor, percentage power loss and power [16] P. R. Satpathy and R. Sharma, ‘‘Power and mismatch losses mitigation by a
enhancement. The algorithms results have been compared fixed electrical reconfiguration technique for partially shaded photovoltaic
with the previously published techniques, such as TCT con- arrays,’’ Energy Convers. Manage., vol. 192, pp. 52–70, Jul. 2019.
[17] M. Horoufiany and R. Ghandehari, ‘‘Optimal fixed reconfiguration scheme
figuration, CS, and GA. In addition, the energy produced and for PV arrays power enhancement under mutual shading conditions,’’ IET
revenue generated for a period of 1 year is calculated and Renew. Power Gener., vol. 11, no. 11, pp. 1456–1463, Sep. 2017.

VOLUME 8, 2020 109623

 T. S. Babu et al.: PV Array Reconfiguration System for Maximizing the Harvested Power

[18] P. R. Satpathy, R. Sharma, and S. Dash, ‘‘An efficient SD-PAR technique THANIKANTI SUDHAKAR BABU (Member,
for maximum power generation from modules of partially shaded PV IEEE) received the B.Tech. degree from Jawahar-
arrays,’’ Energy, vol. 175, pp. 182–194, May 2019. lal Nehru Technological University, Ananthapur,
[19] S. N. Deshkar, S. B. Dhale, J. S. Mukherjee, T. S. Babu, and N. Rajasekar, India, in 2009, the M.Tech. degree in power elec-
‘‘Solar PV array reconfiguration under partial shading conditions for max- tronics and industrial drives from Anna University,
imum power extraction using genetic algorithm,’’ Renew. Sustain. Energy Chennai, India, in 2011, and the Ph.D. degree from
Rev., vol. 43, pp. 102–110, Mar. 2015. VIT University, Vellore, India, in 2017.
[20] P. L. Carotenuto, A. Della Cioppa, A. Marcelli, and G. Spagnuolo,
He is currently working as a Postdoctoral
‘‘An evolutionary approach to the dynamical reconfiguration of photo-
Researcher with the Department of Electrical
voltaic fields,’’ Neurocomputing, vol. 170, pp. 393–405, Dec. 2015.
[21] M. M. Shamseldein and M. Kazerani, ‘‘Reconfigurable photovoltaic struc- Power Engineering, Institute of Power Engineer-
ture,’’ U.S. Patent 13 354 876, Jan. 20, 2012, pp. 1–29. ing, Universiti Tenaga Nasional (UNITEN), Malaysia. He has published
[22] K. Ş. Parlak, ‘‘PV array reconfiguration method under partial shading more than 30 research articles in various reputed international journals.
conditions,’’ Int. J. Electr. Power Energy Syst., vol. 63, pp. 713–721, Acting as a editorial board member and reviewer for various reputed journals
Dec. 2014. like the IEEE, IEEE ACCESS, IET, Elsevier, and Taylor and Francis. His
[23] J. Storey, P. R. Wilson, and D. Bagnall, ‘‘The optimized-string dynamic research interests include design and implementation of solar PV systems,
photovoltaic array,’’ IEEE Trans. Power Electron., vol. 29, no. 4, renewable energy resources, power management for hybrid energy systems,
pp. 1768–1776, Apr. 2014. fuel cell technologies, electric vehicle, and smart grid.
[24] D. Yousri, S. B. Thanikanti, D. Allam, V. K. Ramachandaramurthy, and
M. B. Eteiba, ‘‘Fractional chaotic ensemble particle swarm optimizer
for identifying the single, double, and three diode photovoltaic models’ DALIA YOUSRI received the B.Tech. (Hons.) and
parameters,’’ Energy, vol. 195, Mar. 2020, Art. no. 116979. M.Tech. degrees in electric power and machine
[25] D. Yousri, M. A. Elaziz, and S. Mirjalili, ‘‘Fractional-order calculus- from the Faculty of Engineering, Fayoum Uni-
based flower pollination algorithm with local search for global optimiza- versity, Egypt, in 2011 and 2016, respectively.
tion and image segmentation,’’ Knowl.-Based Syst., vol. 197, Jun. 2020, She is currently pursuing the Ph.D. degree. She
Art. no. 105889. is also working as an Assistant Lecturer. She
[26] D. Yousri, D. Allam, M. B. Eteiba, and P. N. Suganthan, ‘‘Chaotic has published refereed manuscripts in the fields
heterogeneous comprehensive learning particle swarm optimizer vari- of optimization algorithms, photovoltaic applica-
ants for permanent magnet synchronous motor models parameters tions, chaotic systems, and fractional calculus with
estimation,’’ Iranian J. Sci. Technol., Trans. Electr. Eng., pp. 1–20, some topics. Acting as a reviewer for various
Dec. 2019. reputed journals, such as IEEE ACCESS, IET, Energy Conversion and Man-
[27] T. S. Babu, J. P. Ram, T. Dragicevic, M. Miyatake, F. Blaabjerg,
agement, (Elsevier), Applied Soft Computing, and the International Journal
and N. Rajasekar, ‘‘Particle swarm optimization based solar PV array
of Electronics and Communications. Her research interests include the mod-
reconfiguration of the maximum power extraction under partial shad-
ing conditions,’’ IEEE Trans. Sustain. Energy, vol. 9, no. 1, pp. 74–85, ifications of optimization algorithms, modeling and implementation of solar
Jan. 2018. PV systems, chaotic systems, and fractional calculus topics.
[28] A. Fathy, ‘‘Recent meta-heuristic grasshopper optimization algorithm
for optimal reconfiguration of partially shaded PV array,’’ Sol. Energy,
KARTHIK BALASUBRAMANIAN (Member,
vol. 171, pp. 638–651, Sep. 2018.
[29] D. Yousri, D. Allam, and M. B. Eteiba, ‘‘Optimal photovoltaic array
IEEE) received the B.Tech. degree in electrical and
reconfiguration for alleviating the partial shading influence based on a electronics engineering with a minor in industrial
modified harris hawks optimizer,’’ Energy Convers. Manage., vol. 206, automation from the Vellore Institute of Tech-
Feb. 2020, Art. no. 112470. nology (VIT), Vellore, India, in 2014, and the
[30] M. Azharuddin, T. S. Babu, N. Bilakanti, and N. Rajasekar, ‘‘A nearly M.Sc. degree in power engineering from Nanyang
accurate solar photovoltaic emulator using a dSPACE controller for real- Technological University (NTU), Singapore.
time control,’’ Electric Power Compon. Syst., vol. 44, no. 7, pp. 774–782, Prior to pursuing his master’s degree, he was
Apr. 2016. working as an Electrical, Instrumentation Con-
[31] K. Sangeetha, T. S. Babu, and N. Rajasekar, ‘‘Fireworks algorithm-based trol and Telecommunication Engineer at Saipem,
maximum power point tracking for uniform irradiation as well as under Chennai, India, from July 2014 to May 2017. Prior to joining OTD, he was
partial shading condition,’’ in Artificial Intelligence and Evolutionary working at Micron Semiconductor Asia Operations Private Ltd., Singapore,
Computations in Engineering Systems. New Delhi, India: Springer, 2016, as a Process and Equipment Engineer, from August 2018 to November
pp. 79–88. 2019. He is currently working as an Electrical/Automation Engineer with
[32] D. H. Wolpert and W. G. Macready, ‘‘No free lunch theorems for optimiza-
Offshore Technology Development (OTD) Private Ltd., Keppel Offshore
tion,’’ IEEE Trans. Evol. Comput., vol. 1, no. 1, pp. 67–82, Apr. 1997.
and Marine Ltd., Singapore. He has published numerous research papers
[33] R. V. Rao, ‘‘Rao algorithms: Three metaphor-less simple algorithms for
solving optimization problems,’’ Int. J. Ind. Eng. Computations, vol. 11,
in highly reputed national and international journals. He has also served as a
no. 1, pp. 107–130, 2020. reviewer for various reputed national and international journals. His research
[34] S. Balochian and H. Baloochian, ‘‘Social mimic optimization algorithm interests are fuel cell system design, PV system design, power systems,
and engineering applications,’’ Expert Syst. Appl., vol. 134, pp. 178–191, automation, IoT, semiconductors and applications of power electronics
Nov. 2019. in renewable energy systems. He has received multiple high-performance
[35] M. Tahani and N. Babayan, ‘‘Flow regime algorithm (FRA): A physics- leadership awards from the Toastmasters International, USA. He was the
based meta-heuristics algorithm,’’ Knowl. Inf. Syst., vol. 60, no. 2, President of the Toastmasters International—VIT University Chapter.
pp. 1001–1038, Aug. 2019.

109624 VOLUME 8, 2020