Renewable and Sustainable Energy Reviews 53 (2016) 1500–1507

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

journal homepage: www.elsevier.com/locate/rser

A review of maximum power point tracking methods of PV power

system at uniform and partial shading
Liqun Liu a,n, Xiaoli Meng b, Chunxia Liu c
a
Department of Electronic And Information, Taiyuan University of Science & Technology, 030024 Taiyuan, Shanxi Province, China
b
College of Engineering, Xi'an International University, Xi'an, China
c
College of Computer Science and Technology, Taiyuan University of Science & Technology, 030024 Taiyuan, Shanxi Province, China

art ic l e i nf o a b s t r a c t

Article history: The maximum power point tracking (MPPT) methods for the photovoltaic (PV) power system have been con-
Received 26 January 2015 sidered by the experts in the whole world. A review of different MPPT methods is discussed in this paper. Firstly,
Received in revised form the main methods at uniform radiation that will be deliberated are Hill-Climbing (HC), Incremental Conductance
20 May 2015
(IncCond), Perturbation and Observation (P&O), Fuzzy-Logic (FL), and Neural Network (NN). However, these
Accepted 17 September 2015
existing methods have several drawbacks such as slow tracking speed, low tracking efficiency, which fail to
Available online 10 November 2015
extract maximum power at complex partial shading. Then, the more recent MPPT methods at partial shading are
Keywords: discussed to find the best MPPT control strategy, such as the particle swarm optimization (PSO) algorithm, Cuckoo
Photovoltaic (PV) Search (CS) method, and Fibonacci line search scheme, etc. The advantages and shortcomings of abovementioned
Maximum power point tracking (MPPT)
MPPT methods are compared to find the optimal MPPT during partial shading conditions. It is imagined that this
Uniform irradiation
article will be a source of valuable information for PV professionals to keep abreast with the latest progress in the
Partial shading
PV power area, as well as for new researchers to get started on MPPT.
& 2015 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1501
2. General block diagram of PV power system. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1501
3. MPPT methods at uniform radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1502
3.1. The conventional method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1502
3.2. The soft computing method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1503
3.3. Other technologies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1504
4. MPPT methods at partial shading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1504
4.1. Modified the conventional method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1504
4.2. Smart technique. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1504
4.2.1. Particle swarm optimization (PSO) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1504
4.2.2. Cuckoos' behavior method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1505
4.2.3. ANN technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1505
4.2.4. Fibonacci line search scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1505
4.3. Summarize . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1505
5. Conclusiones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1506
Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1506
Appendix A. Supplementary material. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1506
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1506

n
Correspondence to: College of Electronics Information Engineering, Taiyuan University of Science & Technology, 030024 Taiyuan, Shanxi Province, China.
Tel.: þ 86 351 6111696.
E-mail address: [email protected] (L. Liu).

http://dx.doi.org/10.1016/j.rser.2015.09.065
1364-0321/& 2015 Elsevier Ltd. All rights reserved.
 L. Liu et al. / Renewable and Sustainable Energy Reviews 53 (2016) 1500–1507 1501

1. Introduction In a short span of time, numerous conventional MPPT methods

have been proposed, such as the Perturbation and Observation
Renewable energy is free, clean and endless, such as wind [23] or Hill-Climbing [24], Incremental Conductance [25,26],
energy, tidal energy, solar thermal, photovoltaic power, etc., and Fuzzy-logic [27], Artificial Neural Network (ANN) [28,29], Voltage-
the application of renewable resource has been regarded by the based [30], Current-based [31,32], etc. All MPPT control schemes
most of government in the world. Recently, the importance of PV follow the same goal which is extracting as much as possible
power system in the generation of electricity has attracted the electric energy from PV array by using direct current converts into
research attention due to solar energy is a clean, efficient, and direct current converter. However, despite all the aforementioned
environmental renewable and sustainable energy resources [1–8]. method have been successfully demonstrated in tracking the MPP
Especially after the energy crisis, numerous environmental pro- at the uniform radiation where only one MPP exists in the P–V
blems have affected the sustainable development of whole human curve, unfortunately, which is abortively to identify the GP at
beings such as greenhouse gas emissions, acid rain, soil pollution, partial shading where there are multiple MPPs in the P–V curve,
and biodiversity vanishing, etc. At present, the electric power that is, which is unsuccessful in distinguish the GP among the local
generated from solar energy has become one of the major chal- MPPs. In addition, many researches have been carried out to
lenges for engineers and scientists. PV power systems are expected address the partial shading effect by improving the traditional
to play a significant role in future electricity supply and lose reli- MPPT or by using the intelligent algorithms, such as artificial
ance on conventional electricity generated by burning coal, natural neural network [33], particle swarm optimization algorithm
gas, and oil. In a word, the endless solar energy is a unique pro- [34,35], and Fuzzy logic [36]. In a world, these methods include
spective solution for the energy crisis and global warming [9,10]. development of modified MPPT techniques, array reconfigurations,
Colossal amount of work has been carried out to improve the and use of different converter topologies [37].
solar cell design and its fabrication technologies [11–18]. Whilst At present, the technological advance related to this subject has
these efforts are significant, it is crucial to continually improve the grown tremendously. As a conclusion, now is the time to con-
overall performance at the system level. One area of enormous tribute a comprehensive and up to date review of MPPT under
interest is the MPPT. The aim of MPPT is to ensure that at either uniform or non-uniform conditions. In this paper, the conventional
irradiance or temperature, the maximum power is extracted from method, intelligent method, latest work, and the advantages and
the PV modules by matching its power–voltage (P–V) working shortcomings of various MPPT control methods are discussed to
help for certain group of readers in the PV power area, as well as
point with the corresponding power converter [19], which has
for new researchers to get started on MPPT. Moreover, due to the
been and is still attracting immense interest from PV research
number of articles related to these particular methods is very
communities as well as industrial players because it is the most
gigantic, only works with significant contributions are introduced
economical way to improve the overall PV system efficiency [20].
in this paper, it is imperative that these previous works with
Because of the non-linear output characteristics of the PV array,
diminutive modifications or improvements may not be revealed in
the tracking of the MPP at various environmental conditions can
the reference list.
sometimes be a challenging task. The issue becomes more per-
plexing when the whole PV array does not obtain uniform irra-
diance, namely, the partial shading [21]. This issue has received
2. General block diagram of PV power system
much attention lately, mainly due to the actual large scale PV array
is inevitably partially shaded by the passing cloud, shading of
In general, the PV power system includes two types, namely,
building, dust, and excreta of passing bird, partly due to unbefit-
the distributed PV power (DG) system and the connected-grid
ting physical constraints, such as space limitations and unforeseen
(CG) PV power system. The installed capacity of the DG system is
structural development, which strikes on certain spots of the solar
small to compare with that of CG system. DG system is generally
array, while other parts are left uniformly irradiated [20]. The P–V
existed in remote region such as island, remote village where the
curves of a PV array display multiple peaks with several local and conventional power grid is difficult to supply, micro-grid, etc. In
one global peak (GP) at partial shading. Conclusion, how to find which, the storage battery group is necessary to afford the
the only GP is the most challenging task. approving power supply. With the rapid development of PV
In order to improve the efficiency of PV power generation industry, the CG system has become the dominated status, which
system, the MPPT control method is usually an essential part of a has a share of 80% of the total application amount of PV modules.
PV power system because of the non-linear characteristics of the Therefore, the CG system is mainly discussed in this section, which
PV array. Furthermore, it is necessity to design an efficient power consists of PV array, MPPT controller, connected-grid inverter, and
converter that is not only high in efficiency but also optimizes the sensors. In addition, the storage battery group can be omitted in a
output power of the solar cells [22]. It is well known that MPPT is large capacity connected-grid PV generation system. Here, the
basically an operating point matching between the PV modules direct current converts into DC circuit and the DC convert into the
and power converter. However, due to the non-linear current– alternating current (AC) circuit are called the DC/DC converter and
voltage (I–V) characteristics of the PV output curve and the con- DC/AC converter, respectively.
sequence under the varying insolation conditions. Moreover, the A PV array in a typical series-parallel configuration with six PV
output efficiency of solar cell depends on many factors such as modules can be seen from Fig. 1, in which two types of diodes
internal factors (e.g. the diode factor, the series resistance, the (bypass and blocking) are connected. In detail, the PV modules are
parallel resistance, etc.) and external factors (e.g. temperature, connected in strings, with three modules per string. During partial
radiation, dirt, shadow, and so on). An efficient MPPT scheme is shading, several modules in the string are less illumination, its
necessity, which can improve the tracking speed and output effi- voltage drops, and which behave as a load instead of a generator
ciency of power converter. So the tracking of the optimal max- [38–40]. Partial shading reduces the total power generation from
imum power point (MPP) can sometimes be a challenging task to PV array and may cause hotspot problem [41]. In order to protect
extract as much as possible power from the costly PV modules. the shaded module from the hotspot problem, lots of bypass diode
Moreover, the tracking becomes more complicated when the are connected in parallel with a group of PV cells in each PV
entire PV array does not receive uniform radiation—a condition module. In addition, blocking diodes are connected at the end of
known as partial shading [2]. each string to protect the array from being affected by the reverse
1502 L. Liu et al. / Renewable and Sustainable Energy Reviews 53 (2016) 1500–1507

Grid
PV Array
+
IPV

IAC
DC/DC DC/AC
Converter Converter
VDC VAC
VPV

Optimal Optimal
Vref_DC
PWM Connected PWM
MPPT Driver Driver
Grid Control
PLL

Controller Inverter

Fig. 1. A typical connected-grid PV power system.

of PV modules because of simplicity, ease of implementation, few

3000 Uniform
measured parameters, and low cost.
Partial shading 1
Partial shading 2
Certainly, the P&O and HC are not same, and the main different
Output Power (W)

2000
is that the HC method is perturbed in duty cycle of the power
converter and the P&O method is perturbed in the operating
voltage/current of the PV array [2]. The P&O/HC methods are the
B
1000 most commonly used algorithms in partial PV power system.
A
However, these methods have four major drawbacks: (1) slowing
tracking speed at rapid weather change. (2) The oscillation around
0
200 30 the MPP of PV output power cannot reduced, which consequently
100 20
Outp
ut Vo 0 0
10 waste the energy from PV array. (3) They fail to track MPP under
ltage nt (A)
(V t Curre the radiation varies quickly or partial shading conditions. (4) They
) Outpu
cannot compare the array terminal voltage with the actual MPP
Fig. 2. The I–V curves of PV at uniform and partial shading.
voltage, and which cannot track the actual MPP at that time.
In this method, the DC/DC controller adjusts the voltage or
current between the strings caused by the voltage mismatch
current by a small amount from the PV array and computes power;
between the parallel connected strings [20,42].
if the power decreases, further adjustments in the direction are
In normal condition, that is, when the entire PV array gains a
changed, or else, if the power increases, further adjustments in
uniform radiance, as shown in Fig. 2, the P–V cure of PV array exhibits
that direction are held till the power reach the maximum value.
a single MPP, and all bypass diodes of every string are reverse biased.
IncCond method is widely used to track the accurately MPP by
Consequently, the MPPT control method is easy during uniform. The
comparing the slope dI/dV, which is based on the dI/dV¼ 0 at the
P–V cure of PV array, however, depicts several local peaks (LPs) under
MPP, positive on the left of the MPP, and negative on the right. The
partially shaded conditions, and, there is only on global peak (GP), as
equations of IncCond method can be expressed as follows [23]:
can be seen from Fig. 2. During partial shading, as a consequence the
MPPT control strategy becomes a complicated task due to how to dI
¼ 0 at MPP ð1Þ
distinguish the GP from multiple LPs is a difficult assignment. When dV
the bypass diode is removed, the PV array has only a single peak. This
has been discussed in many literatures, but it is realized a major dI
4 0 left of MPP ð2Þ
reduction in output power. Furthermore, it can be confirmed that at dV
partial shading, to improve the output power, the bypass diode must
always connected in parallel with a group of PV cells in each PV dI
o 0 right of MPP ð3Þ
module [43–47]. dV

here I and V are the output current and voltage of the PV array,
respectively. This method has many advantages such as no oscil-
3. MPPT methods at uniform radiation lation occurs around the MPP in steady state. There are three
major drawbacks: (1) the design of controller is complicated, high
3.1. The conventional method cost, and need accurately measure parameters. (2) Slowing track-
ing speed at rapid weather change. (3) It fails to track MPP under
Normally, these MPPT methods at uniform radiation are sys- the radiation varies quickly or partial shading conditions.
tematically categorized into two main groups. First, which is by far The second drawback of P&O or HC method is greatly reduced
the most popular, namely, the conventional method, due to the by using the IncCond method. However, the first and the third
MPP can be gains under uniform radiation conditions. Among drawbacks remain no matter using the fixed-step-size or variable-
those MPPT methods, the P&O and HC method is plain to gain MPP step-size. It is difficult to track the actual MPP at that time.
 L. Liu et al. / Renewable and Sustainable Energy Reviews 53 (2016) 1500–1507 1503

3.2. The soft computing method duty ratio, and the value of PWMs can be chosen as 1/2.
PWM ¼ PWMs þ ΔD ð6Þ
The second group is based on the soft computing technologies.
They are becoming more commonly lately because of the avail- Therefore, the fuzzy rules require 25 fuzzy control rules; these
ability of vast and cheap computing power. Some special power rules are based on the rule table. As an example control rule in
converters are designed to track MPP of PV array which uses the FL Table 1: if E is PB and ΔE is Z then ΔD is PB.
or ANN MPPT method. After fuzzy implication, the FL controller output need convert
Normally, FL control consists of three parts: fuzzification, from a linguistic variable to a numerical variable, a defuzzification
inference method, defuzzification. During fuzzification, numerical stage is needed. In the defuzzification stage, the FL controller
input variables from the control object are converted into lin- output is converted from a linguistic variable to a numerical
guistic variables using several fuzzy levels such three, five, seven, variable, and Fig. 3(b) shows the membership function of output
etc. And the five fuzzy levels is selected in this paper, which is variable ΔD. ΔD can decide the change of the duty ratio of the
switch of a boost converter. The range of each membership func-
called negative big (NB), negative small (NS), zero (Z), positive
tion is decided by the previous knowledge of the author.
small (PS), and positive big (PB). Variables E and ΔE are selected as
FL MPPT method has an excellent performance at varying
input variables, where E is the error which can be expressed as (4).
weather and shows better tracking performance than the P&O or
ΔE is the change in power as expressed in Eq. (5), at a sampling
HC controller [32]. The MPP is tracked by computing the slope of
instant n. Where P (n) and V (n) are the output power and voltage
the power–current characteristic and the slope change. The main
of the PV array at the sampling instant, respectively. There dP=dV
advantages of this method is the rapid tracking speed, ease of
vanishes or dP=dV ¼ 0 at the MPP of a PV array. Fig. 3(a) shows the
implementation, few measured parameters, etc. The disadvantages
membership functions of variables error and change in power.
of the FL method, is that the tracking performance and output
PðnÞ  Pðn 1Þ efficiency are highly dependent on the engineer's technical
EðnÞ ¼ ð4Þ
VðnÞ  Vðn  1Þ knowledge, and the operating duty cycle moves away from the
duty cycle of actual MPP no matter the radiation changes or the
ΔEðnÞ ¼ PðnÞ Pðn  1Þ ð5Þ steady state, since the slope change is less than the designed value,
and the duty cycle variation is neglected by computing the slope
Inference method consists of two parts: fuzzy rule table and and comparing with the rule-based table.
fuzzy rules. The rule table of FL controller can be seen from Table 1. Recently, the application of Artificial Neural Network has
If, for example, the operating point is from the left of the MPP, that entered the PV power field as an estimation method since the high
is E is PB, and ΔE is PB, then ΔD should be PB, that is the duty ratio pattern recognition ability. It has many layers: input, hidden and
should largely increasing, as can be seen in Table 1. The duty ratio output layers. Fig. 4 depicts the normal structure of ANN. For
PWM of switch can be expressed as (6). Here, PWMs is the original MPPT, the input variable to the ANN can be PV array parameters
like PV voltages and currents, environmental data like irradiance
and temperature, or any combination of these. The output is
usually one or several reference signal(s) such as the optimal
voltage ðV mppt Þ, optimal current ðI mppt Þ, and duty cycle signal. In
order to extract the maximum power, the weights associated with
neurons come from a complicated training process based on the
experimental measurements and simulation results. If the training
process is end, the ANN can be used to estimate the MPP of a PV
under uniform radiation conditions.
The advantages of the ANN MPPT method include no require-
ment for knowledge on internal parameters, require less compu-
tational amount, and provide a compact solution for the control
equipment [32]. The ANN MPPT converter has better performance
and effectiveness than that of P&O MPPT converter. Certainly, the
ANN MPPT scheme must be trained by observing different radia-
tion and MPP distribution conditions.
The drawbacks of the ANN MPPT method is similar with FL
MPPT method, is highly dependent on the technical knowledge of
specialist and reliance on the accurate characteristic of the PV

Input Layer Hidden Layer Output Layer

Fig. 3. Membership functions for (a) input E (n) and ΔE. (b) Output ΔD.
Temperature
Imppt
Table 1
Form of the rules of the fuzzy control.

ΔE E j Vmppt
NB NS Z PS PB

NB NB NS NB PS PB
NS NB NS NS PS PB
Wij
Z NB NS Z PS PB
i
PS NB NS PS PS PB
PB NB NS PB PS PB
Fig. 4. ANN structure.
1504 L. Liu et al. / Renewable and Sustainable Energy Reviews 53 (2016) 1500–1507

module that change with time, temperature, radiation, and need two independent control loops to track the MPP. The first control
periodically trained to gain accurate MPP. That is to say, the loop contains the MPPT scheme, and the second one is a propor-
tracking speed is slow and difficult to track the operating point. tional (P) or P-integral (PI) control loop [30]. However, the afore-
mentioned MPPT methods at partial shading are difficult to track
3.3. Other technologies the actual optimal power point at complex partial shading. And
those methods have common faults such as slow tracking speed,
There are other tracking methods such as the linear approx- need training, and getting into the local optimum, etc.
imation method (LA), optimal voltage method (OV), and combined In this section, the performance of three types likes the mod-
method. The LA method uses an important characteristic which is ification of conventional MPPT (P&O, IncCond, and FLC) and smart
that the proportion between the short-circuit current and optimal techniques (PSO, Fibonacci line search method, and Cuckoos'
current is about 0.9 [48]. The OV method uses that the ratio behaviour method) under partial shading are reviewed.
between the open-circuit voltage and optimal voltage is about 0.76
[49]. In addition, the combined method consists of two types: the 4.1. Modified the conventional method
improvement of conventional method, the combination of two or
several method. For example, the traditional P&O method is fixed For partial shading, in Fig. 2 there are several peaks labeled by
step, which is improved by using the variable step method to A and B emerge, clearly, peak B is the global MPP while the others
reduce the oscillation near the MPP. Furthermore, the ANN is named local MPPs. When shading occurs, the conventional MPPT
combined with the IC method as introduced in literature [50]. methods are difficult to identify the global peak from several
peaks, such as P&O, IncCond, and FLC, and so on. Hence, the
remedial effort is necessary to distinguish the GP.
4. MPPT methods at partial shading A two-mode modified P&O is employed in the literature [51–
53], which implements a two-stage method. In the first stage, the
Many MPPT methods have been effectively used in the stand- normal program is used, and in the second stage, the GP tracking
alone and grid-connected PV power system at slow and smoothly method is proposed, that is, the conventional MPPT strategy is
changing radiation. However, it is difficult to directly apply these used to gain the first LP labeled by A as shown in Fig. 2, then, the
approaches into the MPPT of the PV power system at partial adjacent is scanned to gain the GP named B. Certainly, to avoid
shading. In practical applications, a PV array consists of many PV scanning the total P–V curve, many special methods are applied to
modules connected in series and in parallel to provide the reduce the scanning range such as limited two boundary condi-
necessary voltage and/or current. The output power of a PV array tions (Vmin and Vmax), two instantaneous power values compar-
decreased considerably due to soiling, cell damaging, partial ison, a voltage sweep method, etc.
shading, etc. The output power of a PV array has a very significant In [54] firstly, the output voltage and current of PV array are
reduction [26]. Tracking the MPP is not easy when a PV array is forced to operate into 0.8 and 0.9, respectively. Due to the pro-
partially shaded cue to two or more local peaks (LP) may appear. portion between optimal voltage and open-circuit voltage is about
The PV power–voltage–current characteristic shows multi-local 0.8 as compare with that between optimal current and short-
peaks only one is the global peak (GP). The conventional MPPT circuit current is about 0.9. Secondly, the traditional IC method is
cannot distinguish between the GP and LP. There are needs for the employed to acquire the GP. In [55] a similar two-stage IC strategy
special schemes which can distinguish the GP under the condi- is used to compare with that of P&O method. The total algorithm is
tions of varying radiation, temperature or fast moving clouds. divided into two parts: (1) the normal program and (2) the GP
A few papers have been reported trying to avoid this difficulty. tracking method.
Furthermore, most of these control methods are rather compli- A modified FLC method is proposed in literature [36]. The
cated, difficult of implementation, high cost, and require more complete algorithm consists of two parts: (1) the FLC method and
measured parameters 1. Ahmed and Miyatake [22] proposed a (2) the scanning method. Despite this simplification, for partial
simple MPPT method using the Fibonacci line search. The basic shading conditions, the algorithm still needs to scan more than
idea of Ref. [20] is the Fibonacci search algorithm, and this method 60% of the P–V characteristic curve to ensure none of the potential
iteratively restricts and shifts the searching range so as to contain peak GP is missed. In addition, the oscillation of output power is
the optimal point in the searching range. This method can track big during scanning.
the MPP at uniform radiation and slow changing radiation. How-
ever, it is difficult to tract the GP under rapidly changing radiation 4.2. Smart technique
or partial shading conditions. Syafaruddin et al. [32] proposed a
MPPT method for the partially shaded PV array using ANN with 4.2.1. Particle swarm optimization (PSO)
three layer feed-forward with polar information controller. How- PSO is a search method based on a population stochastic
ever, this approach might result in excessive complexity of the optimization algorithm developed by Eberhart and Kennedy in
control scheme and large computation amount. Ref. [33,34] pro- 1995, which is modeled after the social behaviour of bird flocking.
posed the simulation and hardware implementation of MPPT of a PSO is a typical representative of swarm intelligence algorithm,
partial shading PV array using the particle swarm optimization. which maintains a swarm of individuals (named particles) where
The aim is to realize a power tracking method that can find the GP each particle represents a candidate solution, and the position in
to maximize the output power from the PV array. The PSO method optimum space of the particle shows the optimal solution of the
can track the MPP at different weather, but the output character- problem, and the velocity vector of particle determine the direc-
istic of PV array at that time is necessary to distinguish the GP. The tion of the particles and speed value, and each particle obeys the
output curves of PV array need to convert the function, and then current optimal particle, and search the optimization in the solu-
the PSO algorithm is used to find the global optimal solution, that tion space based on itself flying experience [56–58]. Advantages of
is to say, to find the GP at that time. Therefore the computation the PSO method consist of rapidity and flexibility and robustness,
amount and tracking time are big under rapidly changing radiation and the drawback is that the MPP is easy to fall into the local
or partial shading. All of the methods mentioned above did not optimum which generates the failure to track the GP.
deal with much more complicated non-uniform radiation and PSO methods are envisaged very effective to gain the GP under
partial shading. Furthermore, the conventional MPPT systems use partial shading conditions. Liu et al. [56] utilize the Genetic
 L. Liu et al. / Renewable and Sustainable Energy Reviews 53 (2016) 1500–1507 1505

Algorithm (GA) to modify PSO to track the GP; GA is a kind of consists of 200 input–output pairs [67]. However, the liable
natural selection and genetic principle based on the principle of training data set is a difficult task under partially shaded condi-
effective optimization method, the proposed hybrid genetic algo- tions, and those approaches might result in excessive complexity
rithm and adaptive particle swarm optimization (GAAPSO) algo- of the control scheme and large computation amount.
rithm has faster tracking speed and almost zero steady-state K. Punitha presents a simple hybrid method, the NN combines
oscillation as compare with that of P&O. Authors in [57] for- with IC algorithm for MPPT in PV system, an off-line ANN, trained
mulate a current calculated predicted arithmetic based on the using a back-propagation algorithm is utilized for online estima-
output current of PV array at partial shading; then the conven- tion of reference voltage for the feed-forward loop [68].
tional PSO is used to obtain the GP. In another work, voltage based The effective way to utilize IC is to have prior knowledge of V ref
PSO method [58] is proposed, in which the DC voltage super- value. This will reduce the tracking time of the algorithm. In lit-
position principle is utilized to get the output characteristic of erature [68], under partial shading conditions an artificial neural
voltage of PV array; then the traditional PSO is used to gain the GP. network is used to improve the value of V ref at uniform radiation.
As mention above, the study reveals that the PSO method is more And, V ref ¼ KV oc ð0:73 o K o 0:8Þ, which uses the approximately
robust and rapid in tracking the GP than the conventional method, linear relationship between the open circuit voltage and the
which can work efficiently in gaining the GP at complex partial maximum power point voltage. The ANN can be trained offline
shading. and can then be used in an efficient way in the on-line environ-
ment. The effectiveness of IC MPPT method can be improved by
4.2.2. Cuckoos' behavior method combining with ANN, as mentioned in [68].
Cuckoo Search (CS) is an optimization algorithm, is inspired
from the natural parasitic reproduction strategy of cuckoo birds. It 4.2.4. Fibonacci line search scheme
is observed that several species of cuckoos perform brood para- The Fibonacci search technique belongs to the line search
sitism by laying their eggs in communal nests instead of hatching method which is normally used in the optimization technique of
them in their own, such as the Ani and the Guira. The cuckoo one-variable functions. The method can iteratively restrict and
female seeks the search space to find a suitable nest of host bird to change the searching range to contain the optimal value in the
hatch its eggs. In some cases, the host bird cannot distinguish the range. The direction of the changing is decided by the value of the
imitated eggs lay by the cuckoo female at some specific time, so function at two detective points in the range. In other word, the
that the early hatching cuckoos kill some of the host bird's nestling Fibonacci line search scheme is a method of searching a sorted
or eggs due to the eggs of cuckoo hatch earlier than the host bird's array using a divide and conquer algorithm that reduces the pos-
own, and the nestling of cuckoo can get more food. If the host bird sible locations with the optimal value always lying within that
discovers the cuckoo's eggs, a normal phenomenon is that the host range. The changing direction can be either to the left or to the
bird either throws it away or simply abandons its nest and builds a right. It can continuously reduce the range between the optimal V1
new one elsewhere [59,60]. and V2 and belongs to a mathematical search [69,70]. Initially, two
CS is similar with GA and PSO based on a population algorithm, output voltage values of PV array are tested and the output powers
but it is similar with harmony search in selection procedure. And at these voltages are measured. Then, the searching range is
Lévy flight makes that the randomization is much more efficient, shifted either to the left or to the right according to the Fibonacci
which provide faster convergence. Moreover, the number of search strategy. The new power value at a new voltage point is
parameters for tuning in CS is only two to compare with that of GA continuously gained till the MPP is reached.
and PSO are three parameters or more. In addition, the char-
acteristic of CS is independent on the initialization of samples [60]. 4.3. Summarize
To use CS for tracking MPP of PV at partial shading, appropriate
variables must be selected for the search. Two parameters of CS As mentioned above, many MPPT methods at partial shading
algorithm are the output voltage and step size. The output power using the modified traditional methods and smart methods have
for the new voltage sample is compared with that of old voltage been proposed in several literatures such as modified conventional
sample. If the new sample is more than the old sample, and the method, Fibonacci search, ANN, PSO, and CS. Certainly, the actual
maximum power given by the new simple is selected as the new tracking mode is different as can be seen from Table 2. The
best sample. If the new simple is less than the old sample, the tracking speeds of Fibonacci and ANN are slow to compare with
maximum power is kept. The course continues until all samples that of modified tranditional mothod, PSO, and CS are normal.
have reached the MPP [60–65]. Furthermore, those merits and shortcomings are shown in Table 2.
Here, the main merit of modified method is easily implemented
4.2.3. ANN technology and the main drawback is that the output voltage, current, power
There is a little article using the ANN schemes for several are not stable, and the vibration is big. One shortcoming of other
partially shaded conditions are proposed to gain the GP. A MPPT methods is the excessive amount of calculation.
system is proposed for partially shaded PV array by using ANN and At present, there is not a excellent MPPT method under partial
fuzzy logic with polar information controller. The ANN with three shading conditions. The primary reasons can be divided into two
layer feed-forward is trained to determine the global MPP voltage part: (1) the mathematical model of PV array is not accurate,
under several partially shaded conditions. The global MPP voltage specially, when the PV array is partially shaded and (2) fail to
as a reference voltage is used in the fuzzy logic with polar infor- predict the global peak. In my opinion, the futuer development of
mation controller to gain the required control signal for the power MPPT method at partial shading consists of two points: (1) accu-
converter. Another objective of this study is to determine the rate mathematical model and (2) rapid predict GP. In the first part,
estimated maximum power and energy generation of PV system some effective works have been carried out. For example, the lit-
through the same ANN structure [32]. A three layer feed-forward erature [2] proposed a novel DC voltage superposition principle to
ANN combined with an FL control method is proposed in literature gain the output characteristic of PV array at partial shading. In the
[66], which is tested on different types of solar cells such as mono- second part, the image processing method can be applied to pre-
crystalline silicon, thin-film and amorphous silicon. Fuzzy wavelet dict the area of the GP. In particular, first, the shading of PV array is
network is used to identify the GP of non-crystalline Si PV mod- recorded using video, and then these images is transmitted to the
ules. A data set is used to train and test the ANN network, which computer, and the radiation and area of shading are analyzed by
1506 L. Liu et al. / Renewable and Sustainable Energy Reviews 53 (2016) 1500–1507

using the border detection principle and used in the prediction of

Excessive amount of calculation. And the MPP is easy to

fall into the local optimum which generates the failure
It can be trained offline and used in an efficient Excessive complexity of the control scheme and large

Excessive amount of calculation. Deterioration of the

GP.

The output is not stable and vibration is big.

5. Conclusiones

parameters, good robustness, and rapid tracking convergence speed and solution quality.
In this article, the normal MPPT methods under uniform and

Excessive amount of calculation.

partial shading conditions are discussed to find the optimal
tracking method. Judging by the amount of recent work, we can
conclude that the MPPT is still a very active research area to
computation amount. improve the output efficiency of PV array. For uniform radiation,
the MPPT is very simple due to there is only one MPP. Examples of
to track the GP.
successful schemes have been demonstrated by various works
Shortcomings

described in this paper. However, the existing method has some

drawbacks, specially, when the PV array is partially shaded. As for
the partial shading, there is an increasing trend in smart techni-
ques such as ANN, PSO, and Fibonacci, and so on. These existing
tracking stratagems have modified the tracking performance and
Efficient randomization, fast convergence, less

efficiency of PV array under partially shaded conditions. However,

Good robustness and rapid tracking speed

there are some drawbacks existing in those proposed stratagems,

such as poor output power quality, big vibration, slow tracking
speed, excessive amount of calculation, and easy to track the
way in the on-line environment

failure, etc. In my opinion, the accurate mathematical model and

rapid predict GP method are necessary to realize the efficient
MPPT stratagem, and the image processing can be used to predict
Easily implemented

Easily implemented

the GP. As a conclusion, the MPPT method at partial shading

should become the focus of research in future to improve the
tracking speed and output efficiency of the costly PV power
system.
Merits

speed

Acknowledgment
Tracking

Normal

Normal

Normal

Funding for this work was provided by the financial support

speed

Slow
Slow

of China Postdoctoral Science Foundation (Nos. 2014T70234

and 2013M530895), and Soft science project of Shanxi Province
In the first stage, the normal program is used, and in the second stage,

candidate solution, and the position in optimum space of the particle

(No. 2014041021-3).
A typical swarm intelligence algorithm, each particle represents a
A complicated training process based on the experimental

Appendix A. Supplementary material

CS is similar with PSO based on a population algorithm.
Iteratively restricts and shifts the searching range.

Supplementary data associated with this article can be found in

the online version at http://dx.doi.org/doi:10.1016/j.rser.2015.09.065.
shows the optimal solution of the problem.
Comparison between the different methods during partial shading.

References

[1] Liu L, Liu C. Deliberating on renewable and sustainable energy policies in

China. Renew Sustain Energy Rev 2013;17(1):191–8.
the adjacent is scanned.

[2] Liu L, Sun Z, Liu C. MATLAB-based DC voltage superposition principle to study

the output characteristics on PV array at partial shading. Prz Elektrotech
2012;88(12a):284–6.
measurements.
Tracking mode

[3] Liu L, Liu C, Sun Z, Sun C. A survey of China's low-carbon application practice—
opportunity goes with challenge. Renew Sustain Energy Rev 2011;15(6):2895–
903.
[4] Liu L, Liu Z, Sun Z, Liu C. Degraded output characteristic at atmospheric air
pollution and economy analysis of PV power system: a case study. Prz Elek-
trotech 2012;88(9a):281–4.
[5] Liu L, Liu C, Sun Z, Han R. The development and application practice of neglected
Tracking method at partial

tidal energy in China. Renew Sustain Energy Rev 2011;15(2):1089–97.

Artificial neural network

[6] Liu L, Liu C, Yang G, Zheng X, Wang J. An improved PO and immune MPPT
Modified conventional

method for PV generating system under partially shaded conditions. Prz

Elektrotech 2013;89(1a):235–8.
[7] Liu L, Liu C. Feasibility analyses of hybrid wind-PV-battery power system in
Particle swarm
optimization

Cuckoo Search

Dongwangsha, Shanghai. Prz Elektrotech 2013;89(1a):239–42.

[8] Liu L, Liu C. Techno-economic analysis of off-grid renewable energy power
method
Fibonacci
shading

station: a case study. Prz Elektrotech 2012;88(7a):94–8.

Table 2

[9] Pablo S, Jesus L, Alfredo U, Eugenio G, Luis M. On the testing, characterization,

and evaluation of PV inverters and dynamic MPPT performance under real
varying operating conditions. Prog Photovolt Res Appl 2007;15(6):541–56.
 L. Liu et al. / Renewable and Sustainable Energy Reviews 53 (2016) 1500–1507 1507

[10] Alajmi BN, Ahmed KH, Finney SJ, Williams BW. Fuzzy-logic-control approach of a [40] Mastromauro RA, Liserre M, Dell'Aquila A. Control issues in single-stage
modified hill-climbing method for maximum power point in microgrid standa- photovoltaic systems: MPPT, current and voltage control. IEEE Trans Ind Inf
lone photovoltaic system. IEEE Trans Power Electron 2011;26(4):1022–30. 2012;8(2):241–54.
[11] Kuznetsov IA, Greenfield MJ, Mehta YU, Merchan-Merchan W, Salkar G, [41] Silvestre S, Boronat A, Chouder A. Study of bypass diodes configuration on PV
Saveliev AV. Increasing the solar cell power output by coating with transition modules. Appl Energy 2009;86:1632–40.
metal-oxide nanorods. Appl Energy 2011;88:4218–21. [42] Murtaza A, Chiaberge M, Spertino F, Boero D, Giuseppe MD. A maximum
[12] Li M, Ji X, Li G, Wei S, Li Y, Shi F. Performance study of solar cell arrays based power point tracking technique based on bypass diode mechanism for PV
on a trough concentrating photovoltaic/thermal system. Appl Energy arrays under partial shading. Energy Build 2014;73:13–25.
2011;88:3218–27. [43] Maki A, Valkealahti S. Power losses in long string and parallel-connected short
[13] Han X, Wang Y, Zhu L. Electrical and thermal performance of silicon concentrator strings of series-connected silicon-based photovoltaic modules due to partial
solar cells immersed in dielectric liquids. Appl Energy 2011;88:4481–9. shading conditions. IEEE Trans Energy Convers 2012;27(1):173–83.
[14] Zhang W, Zhu R, Liu B, Ramakrishna S. High-performance hybrid solar cells [44] Paraskevadaki EV, Papathanassiou SA. Evaluation of MPP voltage and powerof
employing metal-free organic dye modified TiO2 as photo electrode. Appl mc-Si PV modules in partial shading conditions. IEEE Trans Energy Convers
Energy 2012;90:305–8. 2011;26(3):923–32.
[15] Li M, Liu Y, Wang H, Shen H. Synthesis of TiO2 sub micro-rings and their [45] Ding K, Bian XG, Liu HH, Peng T. A MATLAB-simulink-based PV modulemodel
application in dye-sensitized solar cell. Apply Energy 2011;88:825–30. and its application under conditions of nonuniform irradiance. IEEE Trans
[16] Qiao L, Wang D, Zuo L, Ye Y, Qian J, Chen H, et al. Localized surface plasmon Energy Convers 2012;27(4):864–72.
resonance enhanced organic solar cell with gold nano spheres. Apply Energy [46] Patel H, Agarwal V. MATLAB-based modeling to study the effects of partial-
2011;88:848–52. shading on PV array characteristics. IEEE Trans Energy Convers 2008;23
[17] Xu HB, Hong RJ, Ai B, Zhuang L, Shen H. Application of phosphorus diffusion (1):302–10.
gettering process on upgraded metallurgical grade Si wafers and solar cells. [47] Ishaque K, Salam Z. A deterministic particle swarm optimization maxi-mum
Apply Energy 2010;87:3425–30. power point tracker for photovoltaic system under partial shading condition.
[18] Chou CS, Guo MG, Liu KH, Chen YS, Chou CS. Preparation of TiO2 particles and IEEE Trans Power Electron 2013;60(8):3195–206.
their applications in the light scattering layer of a dye-sensitized solar cell. [48] Liu L, Wang Z. A rapid MPPT algorithm based on the research of solar cell's diode
Appl Energy 2012;92:223–4. factor and reverse saturation current. WSEAS Trans Syst 2008;7(5):568–79.
[19] Kashif I, Zainal S, Amir S, Muhammad A. A direct control based maximum [49] Liu L, Wang Z. A variable voltage MPPT control method for photovoltaic
power point tracking method for photovoltaic system under partial shading generation system. WSEAS Trans Circuits Syst 2009;8(4):335–49.
conditions using particle swarm optimization algorithm. Appl Energy [50] Dzung PQ, Dinh KL, Lee HH, Phuong LM, Dan Vu NT. The new MPPT algorithm
2012;99:414–22. using ANN-based PV. In: Proceedings of IFOST 2010. 2010. p. 1–6.
[20] Kashif I, Zainal S. A review of maximum power point tracking techniques of PV [51] Patel H, Agarwal V. Maximum power point tracking scheme for PV systems
system for uniform insolation and partial shading condition. Renew Sustain operating under partially shaded conditions. IEEE Trans Ind Electron 2008;55
Energy Rev 2013;19:475–88. (4):1689–98.
[21] Ishaque K, Syafaruddin Salam Z. A comprehensive MATLAB Simulink PV sys- [52] Carannante G, Fraddanno C, Pagano M, Piegari L. Experimental performance of
tem simulator with partial shading capability based on two-diode model. Sol MPPT algorithm for photovoltaic sources subject to inhomogeneous insola-
Energy 2011;85(9):2217–27. tion. IEEE Trans Ind Electron 2009;56(11):4374–80.
[22] Ahmed NA, Miyatake M. A novel maximum power point tracking for photo- [53] Koutroulis E, Blaabjerg F. A new technique for tracking the global maximum
voltaic applications under partially shaded insolation conditions. Electron power point of PV arrays operating under partial-shading conditions. IEEE J
Power Syst Res 2008;78(5):777–84. Photovolt 2012;2(2):184–90.
[23] Femia N, Granozio D, Petrone G, Vitelli M. Predictive & adaptive MPPT perturb [54] Kobayashi K, Takano I, Sawada Y. A study of a two stage maximum power
and observe method. IEEE Trans Aerosp Electron Syst 2007;43(3):934–50. point tracking control of a photovoltaic system under partially shaded inso-
[24] Al-Atrash H, Batarseh I, Rustom K. Effect of measurement noise and bias on hill- lation conditions. Sol Energy Mater Sol Cells 2006;90(18):2975–88.
climbing MPPT algorithms. IEEE Trans Aerosp Electron Syst 2010;46(2):745–60. [55] Young-Hyok J, Doo-Yong J, Jun-Gu K, et al. A real maximum power point
[25] Safari A, Mekhilef S. Simulation and hardware implementation of incremental tracking method for mismatching compensation in PV array under partially
conductance MPPT with direct control method using cuk converter. IEEE Trans shaded conditions. IEEE Trans Power Electron 2011;26(4):1001–9.
Ind Electron 2011;58(4):1154–61. [56] Liu LQ, Liu ZQ, Sun ZY. A novel combined particle swarm optimization and
[26] Wu T, Chang C, Chen Y. A fuzzy-logic-controlled single-stage converter for PV- genetic algorithm MPPT control method for multiple photovoltaic arrays at
powered lighting systems applications. IEEE Trans Ind Electron 2000;47 partial shading. J Energy Resour Technol 2013;135(1):52–6.
(2):287–96. [57] Liu LQ, Liu CX, Gao HM. A novel improved particle swarm optimization
[27] Hiyama T, Kitabayashi K. Neural network based estimation of maximum maximum power point tracking control method for photovoltaic array by
power generation from PV module using environmental information. IEEE using current calculated predicted arithmetic under partially shaded condi-
Trans Energy Convers 1997;12(3):241–6. tions. J Renew Sustain Energy 2013;5(6):063139.
[28] Hiyama T, Kouzuma S, Imakubo T. Identification of optimal operating point of [58] Li H, Liu LQ. A novel PSO MPPT control method based on DC voltage super-
PV modules using neural network for real time maximum power tracking position principle under partially shaded conditions. J Vib Control 2014;20
control. IEEE Trans Energy Convers 1995;10(2):360–7. (9):1356–60.
[29] Dorofte C., Borup U., Blaabjerg F. A combined two-method MPPT control [59] Reynolds AM, Frye MA. Free-flight odor tracking in Drosophila is consistent
scheme for grid-connected photovoltaic systems. In: Proceedings of European with an optimal intermittent scale-free search. PLoS ONE 2007;2(4):e354.
Conference in Power Electronics and Applications. 2005. pp.1–10. [60] Jubaer A, Zainal S. A maximum power point tracking, MPPT for PV system
[30] Mutoh N, Ohno M, Inoue T. A method for MPPT control while searching for using Cuckoo Search with partial shading capability. Appl Energy 2014;119:
parameters corresponding to weather conditions for PV generate systems. 118–130.
IEEE Trans Ind Electron 2006;53(4):1055–65. [61] Tarik L, Aouatif A, Mohammed R. A hybrid mobile object tracker based on the
[31] Mutoh N, Ohno M, Inoue T. A control method to charge series-connected ultra- modified Cuckoo Search algorithm and the Kalman Filter. Pattern Recognit
electric double-layer capacitors suitable for photovoltaic generate systems com- 2014;47(11):3597–613.
bining MPPT control method. IEEE Trans Ind Electron 2007;54(1):374–83. [62] Shlesinger MF, Klafter J. Lévy walks versus Lévy flights. In: Stanley HE,
[32] Syafaruddin, Karatepe E, Hiyama T. Artificial neural network-polar coordi- Ostrowski N, editors. Growth and form. Amsterdam: Martinus Nijhof Pub-
nated fuzzy controller based maximum power point tracking control under lishers; 1986. p. 279–83.
partially shaded conditions. IET Renew Power Gen 2009;3(2):239–53. [63] Yang X, Deb S. Engineering optimisation by cuckoo search. Int J Math Model
[33] Miyatake M, Veerachary M, Toriumi F, Fujii N, Ko H. Maximum power point Numer Optim 2010;1(4):330–43.
tracking of multiple photovoltaic arrays: a PSO approach. IEEE Trans Aerosp [64] Yang XS, Deb S. Multiobjective cuckoo search for design optimization. Comput
Electron Syst 2011;47(1):367–80. Oper Res 2011;40(6):1616–24.
[34] Chowdhury SR, Saha H. Maximum power point tracking of partially shaded [65] Mantegna RN. Fast, accurate algorithm for numerical simulation of Lévy stable
solar photovoltaic arrays. Sol Energy Mater Sol Cells 2010;94(9):1441–7. stochastic processes. Phys Rev E 1994;49(5):4677–83.
[35] Chen Z, Mang X, Pan J. An integrated inverter for a single-phase single-stage [66] Syafaruddin E, Hiyama KT. Artificial neural network-polar coordinated fuzzy
grid-connected PV system based on Z-source. Bull Pol Acad Sci Tech Sci controller based maximum power point tracking control under partially
2007;55(3):263–72. shaded conditions. IET Renew Power Gen 2009;3(2):239–53.
[36] Liu LQ, Liu CX, Wang JA, Kong YG. Simulation and hardware implementation of [67] Syafaruddin E, Hiyama KT. Fuzzy wavelet network identification of optimum
a hill-climbing modified fuzzy-logic for mppt with direct control method operating point of non-crystalline silicon solar cells. Comput Math Appl
using boost converter. J VibControl 2015;21(2):335–42. 2012;63(1):68–82.
[37] Bidyadhar S, Raseswari P. A comparative study on maximum power point [68] Punitha K, Devaraj D, Sakthivel S. Artificial neural network based modified
tracking techniques for photovoltaic power systems. IEEE Trans Sustain incremental conductance algorithm for maximum power point tracking in
Energy 2013;4(1):89–98. photovoltaic system under partial shading conditions. Energy 2013;62:330–40.
[38] Spertino F, Akilimali JS. Are manufacturing I–V mismatch and reverse currents [69] Nabil AA, Masafumi M. A novel maximum power point tracking for photo-
key factors in large photovoltaic arrays? IEEE Trans Ind Electron. 2009;56 voltaic applications under partially shaded insolation conditions. Electr Power
(11):4520–31. Syst Res 2008;78(5):777–84.
[39] Drif M, Perez PJ, Aguilera J, Aguilera JD. A new estimation method of irradiance [70] Ramaprabha R, Balaji M, Mathur BL. Maximum power point tracking of par-
on a partially shaded PV generator in grid-connected photovoltaic systems. tially shaded solar PV system using modified Fibonacci search method with
Renew Energy 2008;33(9):2048–56. fuzzy controller. Electr Power Energy Syst 2012;43(1):754–65.