You are on page 1of 11

Accelerat ing t he world's research.

An Accurate Method for MPPT to


Detect the Partial Shading
Occurrence in PV System
Jubaer Ahmed

IEEE Transactions on Industrial Informatics

Cite this paper Downloaded from Academia.edu 

Get the citation in MLA, APA, or Chicago styles

Related papers Download a PDF Pack of t he best relat ed papers 

An Improved Met hod t o Predict t he Posit ion of Maximum Power Point During Part ial Shading …
Jubaer Ahmed

A Modified P&O Maximum Power Point Tracking Met hod wit h Reduced St eady St at e Oscillat ion and I…
Jubaer Ahmed

Experiment al assessment of maximum power point t racking met hods for phot ovolt aic syst ems Exp…
J-p Barbot , Rachid Boukenoui
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TII.2017.2703079, IEEE
Transactions on Industrial Informatics

An Accurate Method for MPPT to Detect the


Partial Shading Occurrence in PV System
Jubaer Ahmed, Member, IEEE, and Zainal Salam, Member, IEEE

mechanism to differentiate between the local and global peak.


Abstract—This paper proposes an accurate detection scheme Although, numerous intelligent MPPT are suggested in
that effectively differentiates the partial shading from the literature [1], it is surprising to note that most techniques are
uniform change of irradiance. By doing so, it avoids the
unable to detect the occurrence of partial shading reliably and
unnecessary global peak search which results in a drop of the
maximum power point tracker (MPPT) efficiency. The detection consistently [2-9]. If a wrong decision is made—for example a
is achieved by calculating the irradiance at two designated points partial shading is assumed to take place (but actuality it does
on the I-V curve namely, i.e. the short circuit (Isc) and MPP (Impp) not), unnecessary search for a global search is initiated. This
currents. Since the mismatch of irradiance at these two points confusion results in efficiency drop as the operating voltage is
differs greatly for the partial shading and uniform irradiance forced to move at different places along the P-V curve in its
change, the occurrence of the former is easily discriminated. To
effort to locate the global peak. The most common method to
prove its effectiveness, the scheme is integrated into perturb and
observe (P&O) and particle swarm optimization MPPT identify the occurrence of partial shading is to check the
algorithms using a buck-boost converter. Its performance under (sudden and large) change in output power of the PV array,
several partial shading and dynamic shading condition is i.e. ΔP [10-13]. A threshold (Pth) is set for ΔP; if it exceeds a
simulated using Matlab/Simulink and validated using the dSpace certain value for consecutive control samples of the voltage,
DS1104 platform. It only requires 3 samples to determine if the MPPT considers that partial shading occurs. Otherwise, a
partial shading occurs; without the scheme, an unnecessary scans
uniform fluctuation of the irradiance (G) is assumed. Despite
of the entire P-V curve is initiated. Consequently, MPPT
transient efficiency is increased by 30-35%. In addition to this, its simplicity, the checking mechanism is not conclusive. It is
the calculated irradiance is utilized to update the open circuit known that fluctuations of G under non-partial shading can
voltage of the array, thus eliminates the use of temperature and also create the similar change in ΔP. Alternatively, authors in
irradiance sensors. [14] used the current threshold (Ith), instead. However, the
ambiguity to distinguish between these two scenarios remains.
Index Terms—PV, Solar, Partial Shading, tracking MPP, Open Furthermore, none of the published detection method
circuit voltage
specifically indicates how the threshold, (i.e. Pth or Ith) is
determined. Thus, it is impossible to ensure its correct value
I. INTRODUCTION
without resorting to trial and error method. If Pth is too large,

P artial shading in a PV system is a condition whereby only


a certain portion of the PV module (or array) is shaded,
while other parts are remained uniformly irradiated. It is
the MPPT cannot initiate the global peak search, even if it
existed. On the contrary, if the value is too small, the
algorithm is prone a false trigger, hence forcing the
typically caused by shadow from the nearby buildings, trees, unnecessary search for the global peak. Several values of Pth
poles, chimney, overhead cables etc. When the array is are suggested in literature; for example in [15], Pth = 5 W for a
subjected to partial shading, considerable amount of energy is 35 W module [16, 17]. By interpolation, Pth is taken to be
lost because the shaded module is short-circuited by its approximately 15% of rated array capacity. However, this
respective bypass diode, thus forcing its voltage (and power) condition does not guarantee a unique detection for partial
to zero. As a consequence, multi-modal P-V curve, with shading. As argued by [18], it is possible that a 15% change
several local and a global peak is generated. To extract the could be due to the variation of G of 150 W/m2. In [19], the
maximum power from the array, the maximum power point author proposed Pth to be 0.1~0.2 of Pmpp, and in another
tracker (MPPT) algorithm has to initiate a searching recent work [20], Pth is set to 5% of the nominal power. There
This work was supported by the Ministry of Science, Technology and is no evidence to suggest the suitability of this value, but in
Innovation, Malaysia, under Science Fund Grant SF 03-01-06-SF1399. The
practice, it is considered to be too small. In [21], the concept
grant was managed by the Research Management Centre, Universiti Teknologi
Malaysia, under Vote R.J130000.7909.4S119. of normalized deviation of power (ΔP/P) is proposed. The
Jubaer Ahmed is a lecturer in the Faculty of Engineering, Computing and value of Pth is set to 15% of the rated PV power. The same
Science, Swinburne University of Technology Sarawak, 93350, Malaysia. He method is adopted in [10-12, 22, 23], but in [22], Pth is set to
was with Centre of Electrical Energy Systems, Faculty of Electrical
Engineering, Universiti Teknologi Malaysia, Johor Bahru 81310, Malaysia. 10%. Again, no specific reason is given for the selection of
(E-mail: jahmed@swinburne.edu.my). these values.
Zainal Salam (corresponding author) is with the Centre of Electrical Energy In a separate work, the authors in [24-26] imposed both
Systems, Faculty of Electrical Engineering, Universiti Teknologi Malaysia,
Johor Bahru 81310, Malaysia. He is also attached to the Institute of Future current (Ith) and voltage (Vth) thresholds. They kept Ith and Vth
Energy, Universiti Teknologi Malaysia, Johor Bahru 81310, Malaysia. as tuning parameters. Nevertheless, Ith is easily breached
E-mail: zainals@fke.utm.my).
Copyright © 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
1551-3203 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TII.2017.2703079, IEEE
Transactions on Industrial Informatics
2

whenever G changes abruptly. In [27], Pth is assigned as a detection scheme is explained. To quantify its performance,
variable that requires certain tuning under different weather the scheme is integrated with two widely used MPPT, namely
conditions. In addition, Ith and Vth are chosen as 0.1 and 0.2, the adaptive P&O [32] and particle swarm optimization (PSO)
respectively. According to the authors, these two values [21]. They idea is simulated in Matlab/Simulink and
distinguish between power change due to partial shading and implemented in the buck-boost converter using the dSpace
load variation. However, they cannot be used to determine (DS 1104) platform. Several tests are carried out to prove the
whether the power change is due to partial shading or sudden effectiveness of the MPPT algorithm, with and without the
irradiance change—making it prone to false triggering. proposed detection scheme.
Another detection approach is proposed in [28], and later
adopted by [29]. In this method, the MPPT scans three II. THE PROPOSED DETECTION SCHEME OF PARTIAL SHADING
specific points on the P-V curve based on predetermined duty
A. Concept
cycle. Then it imposes Ith and Vth to 0.2 and 0.1, respectively.
Although effective in most of the cases, this method is To appreciate the proposed detection scheme, it is important
unsuccessful under specific partial shading curves, as to first understand the I-V relationship of the PV cell under the
uniform G. For this purpose, the single diode equivalent
explained in [1]. Furthermore, the threshold values need to be
circuit model [31], shown in Fig. 1 is considered.
tuned for specific shading cases. Since this method is based on
the “direct duty cycle” MPPT, therefore cannot be applied in I RS
voltage based types.
A recent technique in [30] proposes a partial shading RP
IPV V
detection by measuring a voltage mismatch in PV modules.
Id
This is straightforward approach because under partial
shading, the voltages of the shaded and unshaded modules
differs significantly. However, to apply this scheme, it is Fig. 1: Single diode model with Rs and Rp
  V  IR    V  IR 
necessary to install one voltage sensor for every module; thus
I I  I exp s 1   s

  VT    R p 
o
making it prohibitively expensive. In another work [31], a

 I pv _ STC  K I T 
pv
relationship is proposed based on the normalized current (1)
Where, V  
change in the vicinity of the MPP. It is observed that the nkT G
and I
change in current is much higher under uniform G, as T q pv GSTC
compared to partial shading. Further, it is assumed that local In (1), Ipv is the photocurrent (in A), I0 is the reverse
peaks are located at integer multiples of 0.8Voc under partial saturation current (in A), Rs is the series resistance (in Ω), Rp
shading. However, this relationship is not necessarily true—as is the parallel resistance (in Ω), n is the diode factor, q is the
argued by [18]. electron charge (in C), k is the Boltzmann’s constant (in
Despite the numerous work carried out to address this issue, J/Kelvin), T is the panel temperature (in Kelvin), KI
there is still a need to find an effective method that can temperature coefficient of current. If the variation of G is
consistently distinguish between partial shading occurrence considered while T remains constant, then Ipv (1) can be
and uniform irradiance change . Thus, this paper proposes an rewritten as


improved detection by scanning two designated points on the I pv G
I-V curve namely, short circuit current (Isc) and the MPP (2)
I pv _ STC GSTC
current (Impp). Afterwards, the irradiance is calculated from the
where, GSTC is the irradiance at standard test condition, i.e.
recorded currents using PV equations. Since the difference in
1000 W/m2 at 25oC. Considering the fact that the maximum
calculated irradiance (mismatch) between these two points
current (Impp) normally lies within the vicinity of 0.9Isc [33],
differs greatly between the partial shading and uniform Eqn. (2) can be approximated as
irradiance change, the occurrence of the former can be easily

 
discriminated. Consequently, the algorithm initiates the
I sc I mpp G
search for global peak (of the multiple peak P-V curve) only
when it is sure that partial shading has taken place. Else, it will I sc _ STC I mpp _ STC GSTC (3)

G  GSTC 
locate the MPP position of the uniform irradiance (which has I sc I mpp
only one unique peak). Since the procedure takes only three GSTC
control samples, the transient efficiency of the MPPT is I sc _ STC I mpp _ STC
significantly improved. In addition, the proposed scheme has The detection concept is outlined by the numerical
another inherent advantage: the open circuit voltage of the computation of the MSX60 module. The specifications of this
array can be determined using the information from the module are provided in TABLE I, while its I-V curve for
calculated irradiances at the two designated points mentioned selected values of G is shown in Fig. 2. By observing Isc, the
above. Consequently, the temperature/irradiance sensors is no drop of G from 1000 to 600 W/m2 results in the following
ratio in the short circuit currents, i.e.
  0.596  0.6
longer necessary, leading to lower implementation cost.
The paper begins with an analysis of PV module behavior I sc 2.265
(4)
under varying G and T. Then the concept of the proposed I sc _ STC 3.8

1551-3203 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TII.2017.2703079, IEEE
Transactions on Industrial Informatics
3

Which is the ratio between the respective values of G. While at I0.8Voc_array the value of G2 is

G2 
Coherently, same conclusion can be drawn from the ratio of I 0.8Voc _ array
the MPP currents when G drops from 1000 to 600 W/m2 i.e. GSTC (8)
I mpp _ STC
  0.596  0.6
I mpp 2.086
 G2  1000  1012
(5) 3.542
I mpp _ STC 3.5
TABLE I
3.5
The specifications of the PV module (MSX 60) used in this paper It can be seen that the values of G1 and G2 are very close to
Parameters Variable Value Unit actual the G i.e. 1000 W/m2. However, it must be noted that
Short Circuit current Isc 3.8 A (6) is approximated based on the observation that Impp is
Open circuit voltage Voc 21.1 V
located at 0.9Isc. This is not necessarily true under all values of
Current at Pmax IMPP 3.5 A
Voltage at Pmax VMPP 17.1 V irradiance. Consequently, it is expected that the computation
Maximum power PMPP 59.85 W of G using (6) inevitably introduces a mismatch value. For
Voc coef. of temperature Kv -0.08 V /oC example, in the above calculation (for G=1000 W/m2), the
Isc coef. of temperature Ki 3e-3 A / oC
mismatch between G1 and G2 is 16.7 W/m2. On the other hand,
for G=600 W/m2, the mismatch is decreased to nearly zero,
X=Voltage while at 300 W/m2 the mismatch increases to 17 W/m2. Based
G=1000 W/m2 Y=Current
on these observations, an important deduction can be made:
Current (A)

once G1 and G2 is calculated and mismatch remains below a


certain threshold, the algorithm treats the condition as a
0.8Voc_array

G=600 W/m2 uniform irradiance.


0.8Voc

To determine that threshold value, an extensive survey is


conducted on I-V curves of 10 monocrystalline and 10
G=300 W/m2 polycrystalline modules, as listed in Table II. The mismatch
between G1 and G2 is plotted vs irradiance in Fig. 3 (a) and
(b). It can be clearly seen that the mismatch for both types
Voltage (V) under any irradiance remains below 40 W/m2. Thus, it can be
concluded that if the mismatch is less than 40 W/m2, uniform
Fig.2: I-V curve to clarify (4) and (5)
irradiance prevails. Based on this conclusion, the following
The proposed detection scheme scans two designated points relationship is applied to differentiate between partial shading
on the I-V curve, namely Isc and Impp. As highlighted by Eqn. and uniform irradiance condition.
| G1  G2 | 40..............Uniform Irradiance
(3), the calculated values of G at the short circuit (Isc) and peak

| G1  G2 | 40..............Partial Shading
(Impp) conditions are almost equal. According to [33], Isc is in (9)
the vicinity of 0.8Voc, while Impp lies near 0.8Voc_array,. The Voc
and Voc_array refer to the open circuit voltage of a single module TABLE II
and array, respectively. Accordingly, in terms of current, Isc The manufacturer and PV module names
can be represented as I0.8Voc and Impp can be denoted as Monocrystalline Polycrystalline
Manufacturer Module Manufacturer Module
I0.8Voc_array. Using these terminologies, (3) can be rewritten as

G GSTC 
Solaria XT400 Kyocera KD145
I 0.8Voc I 0.8Voc _ array Sunmodule SW265 Solarex MSX60
GSTC (6) Sunpower SPR90 Solarex MSX83
I sc _ STC I mpp _ STC
Shell SP150 Aleo Solar P18
When the MPPT acquires samples at 0.8Voc and 0.8Voc_array, it Siemens SM55 Shell S75
records the respective current as I0.8Voc and I0.8Voc_array. From Panasonic N330 Solarland SLP150
these values, G at these two points is calculated using (6). For Mitsubishi MLT265 S-Energy SN60
brevity, the values of G calculated at I0.8Voc and I0.8Voc_array is Hyundai HIS-M300TI Trinal Solar TSM220
LG LG265S1C Yingli YL135
defined as G1 and G2, respectively.
Canadian Solar CS6P220 Zytech ZT300P
B. Detection of Uniform Irradiance Condition
A. Partial Shading Condition
If the array is under uniform irradiance, the values of G1 and
G2 are almost equal because that the slope of the current in the On the other hand, for partial shading, multiple stairs I-V
I-V curve between I0.8Voc and I0.8Voc_array is very small. This fact curve [18, 31], such as shown in Fig. 4 is generated. In this
is numerically verified using the I-V curve in Fig. 2. For three case, the curves are due to the shading at three levels of
values of irradiances, namely 1000, 600 and 300 W/m2, the irradiances, i.e. 1000, 800 and 400 W/m2. Although I0.8Voc
respective current at I0.8Voc and I0.8Voc_array are marked. At STC, remains at Isc (similar to the uniform irradiance case), the
Isc=3.8 A and Impp = 3.5 A. Thus, using (6), the value of G1 at I0.8Voc_array is positioned at a much lower value due to the
I0.8Voc is calculated as continuous drop of the staircase curve. Consequently, the

G1 
values of G1 and G2 differ greatly, as illustrated by the marked
I 0.8Voc
GSTC (7) points of I0.8Voc and I0.8Voc_array. Using (7) and (8), G1 = 996
I sc _ STC W/m2, while G2 = 414 W/m2. The mismatch between the two

 G1  1000  995.3
3.782 is 582 W/m2, which is much larger that the case for uniform
irradiance.
3.8

1551-3203 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TII.2017.2703079, IEEE
Transactions on Industrial Informatics
4

As suggested by [33, 36], the approximate position of the


Irradiance mismatch
MPP voltage Vmpp is 0.8Voc_array at any temperature. While
tracking at MPP, the algorithm is able to update the open
circuit voltage by using (11). As being an approximation
under uniform irradiance, open circuit voltage is represented
as Voc_array,u.

Voc _ array ,u 
Vmpp
(11)
Irradiance (W/m ) 2
0.8
Step 2: when the irradiance changes or partial shading takes
place, the MPPT scans 0.8Voc and 0.8Voc_array voltages to
Irradiance mismatch

determine G1 and G2 respectively. It is evident from the


Section II (B) that the value of G1 is more accurate than G2.
Thus, Voc_array can be updated by modifying the (10) to
 G 
Voc _ array  Voc _ array ,u  aVt N s ln  1 
 GSTC 
(12)
2
Irradiance (W/m )
Fig. 3: Mismatch in calculated G1 and G2 vs. irradiance(a) Monocrystalline Afterward, dividing the Voc_array by number of modules in

Voc  Voc _ array N s


(b) Polycrystalline series (Ns), Voc of a single module is found.
Since the mismatch under partial shading differs by one (13)
order of magnitude, the MPPT can easily distinguish this
condition to the uniform change in irradiance. Similar The scheme can be explained more clearly by using the I-V
observation can be drawn from the Curve 2 which is shaded curve in Fig. 5 where four cases are considered: 1) from 1000
by four levels of irradiances, i.e. 1000, 600, 400 and 200 to 700 W/m2, 2) from 1000 to 400 m2 and 3) from 1000 to 100
W/m2. The calculated values of G1 and G2 is 996 and 217 W/m2 and 4) from 1000 W/m2 to partial shading curve. At
W/m2, giving a mismatch of approximately 780 W/m2 1000 W/m2 and 250 C, the MPP is tracked at 170V. Using
(definitely over 40). (11), the algorithm calculates Voc_array,u as 212.5 V. Note that
the original Voc_array,u is 211 V. Thus the error for this
approximation is 0.7%. If G drops from 1000 to 700 W/m2,
X=Voltage
Y=Current
the MPPT triggers the partial shading detection scheme
because it senses a large power deviation. So, it scans the
0.8Voc_array and 0.8Voc voltage and calculate G2 and G1
Current (A)

respectively. Afterwards, using the value of G1 at 0.8Voc


(695.79 W/m2), the Voc_array can be calculated using (12)
 695.8 
Voc _ array  212.5  1.2  0.8 10  ln    209 V
Curve 1

 1000 
Curve 2

Voltage (V)
Fig. 4: The I-V curve for partial shading

B. Updating the Voc_array


Another advantage of the proposed detection scheme is the
ability to update the value of the open circuit voltage Voc (or
Voc_array for an S-P array) without any sensor. Most of the
voltage-based MPPT techniques [3, 4, 25, 31, 34] use either
temperature or irradiance sensors (or both) to continuously
calculate Voc. This parameter is needed to initialize the
sampling points, to determine the perturbation size and to
predict the local peak position under partial shading. Its value Fig. 5: The I-V curve to clarify Voc_array updating process
can be calculated as [35]
 G 
Voc  Voc _ STC  Kv T  TSTC   aVt ln 
TABLE III


Calculation of Voc_array using (12) for Fig.5

 GSTC 
(10) Change in G1 using Voc_array using Actual Prediction
G (W/m2) Eq. (7) Eq. (12) Voc_array error
Since Voc is G and T dependent, and vary throughout the 1000 to 700 695.79 209.00 207 0.96%
1000 to 400 396.57 203.61 201 1.29%
day, it is necessary to measure the instantaneous values of G 1000 to 100 97.11 190.10 186 2.15%
and T. Usually, the G sensor is ignored because T is a more 1000 to PS 798.94 210.4 204 3.1%
dominant parameter in (10). In the proposed scheme, however, In Fig. 5, the actual value of Voc_array is 207 V. Thus the
the computational of Voc is done by utilizing the information error in the prediction is less than 1%, which can be
from G1 and G2. The process is performed in two steps. considered negligible. The predicted values of Voc_array for all
Step 1: Prior to the occurrence of large irradiance change or three cases are summarized in TABLE III. It is clear from the
partial shading, it is assumed that the MPP is already tracked.

1551-3203 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TII.2017.2703079, IEEE
Transactions on Industrial Informatics
5

table that, the calculated Voc_array is very close the actual IV. RESULTS & DISCUSSION
Voc_array. It is noticeable that under lower irradiance levels or The detection scheme is incorporated into the P&O [32] and
under partial shading deviation has increased. Nevertheless, particle swarm optimization (PSO) MPPT [21]. The former is
error margin is around 1~3%, which is considerably low. The the improved version of the adaptive P&O which is capable of
open circuit voltage for a single module (Voc) can also be handling partial shading condition. The latter is based on the
calculated by dividing Voc_array with number of modules in one soft computing technique, which effectively scans the P-V
string (Ns), as given in (13). curve in search for the global peak. The algorithm is simulated
using the Matlab/Simulink PV system model, proposed in
III. FLOW CHART OF THE DETECTION SCHEME [37]. For experimental validation, MPPT algorithm is
The flowchart of the detection scheme is presented in Fig. 6. implemented on a 120 W buck-boost converter. The converter
It can be readily integrated into the conventional and soft is designed to operate in the continuous inductor current
computing MPPT by incorporating several lines of codes into mode, with the following specifications: switching frequency
the respective algorithm. After the initialization stage, the (f) = 50 kHz, inductor (L) = 0.3 mH and capacitor C1 = 470 μF
MPPT detects the MPP (Vmpp, Impp) and updates the value of and C2 = 220 μF (details in [38]). The codes are programmed
Voc_array1. Whenever a normalized power change over the using the dSpace 1104 digital signal processing board in
threshold is detected, the proposed partial shading checking conjunction with PV array Simulator (PVAS). The PVAS is a
mechanism will be initiated. Then the MPPT scans two linear dc power supply that provides real time simulation of
specific voltages, i.e. V1=0.8Voc_array and V2=0.8Voc to record any time series of G and T, for any arbitrary I–V curves. Thus
current I0.8Voc_array and I0.8Voc, respectively. Afterwards, using it can be used to emulate the behavior of PV modules in
(7) and (8) G1 and G2 are calculated. In addition to that controlled conditions.
Voc_array and Voc will be calculated by using (12) and (13). Since the PVAS utilizes linear MOSFET at the power stages
Then the mismatch (which is the absolute difference between (as opposed to switched mode MOSFET that is used is most
G1 and G2 is calculated. If the mismatch is greater than 40 simulators) the MPPT algorithm can track the MPP very
W/m2 MPPT decides that partial shading has taken place and precisely, due to the absence of the voltage ripple.
initiate global peak search. Otherwise it reverts to track the Furthermore, it is equipped with digital control that allows
MPP under uniform irradiance condition, i.e. at 0.8Voc_array high precision dynamic MPPT measurements. The overall set-
On another note, I0.8Voc and I0.8Voc_array are located at the up of the hardware implementation is shown in Fig. 7.
beginning and at the end sections of the I‒V curve,
A. Integration with the P&O MPPT
respectively. As a result, for a large S-P array, the voltage
difference between these points can be very large. If the The P&O MPPT in [32] is the adaptive type that has the
MPPT algorithm attempts to jump from 0.8Voc_array to 0.8Voc, ability to locate the global MPP under partial shading. It
the PI controller does not have sufficient time to reach at the achieves this goal by scanning the integer multiples of 0.8Voc
desired voltage level. Thus, another intermediary point (i.e. points on the I-V curve. It is further recommended in [31] that
0.5Voc_array) can be assigned so that the scanning from these temperature sensors is fitted to update the value of Voc
two points can be done in two steps. continuously. To demonstrate the performance under change
in uniform irradiance, a 10 × 2 S-P array is utilized. The value
of G is varied uniformly in four steps, i.e. 500, 1000, 700 and
400 W/m2. The step change is applied at every 1 second
interval. The computed STC values are: Voc_array = 210 V, Isc =
7.6 A, Impp= 7 A and Vmpp = 171 V. Throughout the
temperature is assumed to be constant at 250 C.
Fig. 8 demonstrates the tracking action of the P&O without
the inclusion of the proposed detection scheme. When G
changes from 500 to 1000 W/m2, the algorithm sense a sudden
change in power (ΔP). It assumes a partial shading has
occurred and starts to scan the entire P–V curve. However, it is
known (from the set-up) that the power change is due to
uniform change in irradiance, not partial shading. As a result
of this confusion, erroneously scanning takes place and the
MPPT loses unnecessary power during the search duration.
Similar situation is experienced at 2 s and 3 s. In all these
cases, the energy yield of the system drops significantly
during the transient state.
The tracking behavior of the proposed detection scheme is
shown in Fig. 9. For clarity, VPV and IPV waveforms are
enlarged from 1.02 to 1.09 s. When there is large change in
power, the MPPT jumps to 0.8Voc_array and 0.8Voc. From the
VPV trace, it can be observed that the MPPT locates the
0.8Voc_array and 0.8Voc points at 168 and 16.55 V, respectively.
Fig. 6: Flowchart of the proposed scanning mechanism The corresponding current at these two points are 7.11 and

1551-3203 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TII.2017.2703079, IEEE
Transactions on Industrial Informatics
6

1015 W/m2. Since (G1  G2) < 40, the algorithm has
7.57 A. Thus, according to (7) and (8), G1= 995.4 and G2 = Without
proposed

Efficiency (%)
mechanism
acknowledged that the condition is not partial shading. Thus, it ηavg=55%

updates the Voc using (13) as 210 V. Thereafter it correctly


tracks the MPP at 0.8Voc_array (168 V).
With proposed
mechanism
ηavg=85%

Time (s)
Fig. 10: Efficiency improvement through proposed mechanism
Fig. 11 shows the experimental result that replicates the
simulation. When there is a step change in the G, the P&O
unnecessarily triggers the search for global peak; the
oscillogram in this figure is in agreement with simulation
shown in Fig. 8. On the other hand, when the detection
scheme is integrated into the algorithm, the tracking improves
significantly, as presented in Fig. 12. As expected, it
recognizes that this is not a partial shading case. Thus it
Fig. 7: Hardware implementation setup updates the Voc_array and determines the MPP very quickly.

VPV (V)
Transient
scanning on
VPV (V)

whole PV curve

IPV (A)
IPV (A)

Transient
PPV (W)

power loss

Time (s)
PPV (W)

Fig. 11: Performance of [32] in hardware


VPV (V)

Two point
Time (s) scanning

Fig. 8: Performance of [32] in original version


IPV (A)
VPV (V)

Transient
efficiency
PPV (W)

improved

Time (s)
IPV (A)

Fig. 12: Performance of [32] in hardware integrated with proposed


detection mechanism
The next test is to demonstrate the ability of the proposed
PPV (W)

scheme to detect the partial shading condition consistently.


Four levels of G [1000, 800, 500, 300 W/m2] are used to
create the shading pattern. As expected, the resulting I-V and
Time (s)
P-V curve is characterized by four peaks, as shown in Fig. 13.
The test sequence is designed for 3 seconds. The partial
shading is set to take place at 1 s (for a duration of one
second); then the uniform condition is restored for 1 s. Fig. 14
shows the results of the test when proposed detection scheme
Fig. 9: Performance of [32] integrated with the proposed mechanism is used [32]. When partial shading is imposed, the MPPT
The detection process with the proposed scheme requires scans the 0.8Voc_array and 0.8Voc points and calculate G1 and G2
only 3 samples to locate the MPP. Without the scheme, it using (7) and (8). Since the mismatch between G1 and G2 is
takes 10 samples. Since the latter scans the entire curve, the more than 40, the MPPT confirms that partial shading has
time taken to search for the global peak causes the transient taken place. Then, it initiates the search and locates the global
efficiency (ηavg) drops to 55%. On the other hand, when the peak correctly at 85 V. When uniform irradiance is restored
scheme is implemented, the transient efficiency [39] is kept at after 2 s, the scheme again senses a large power change; thus it
85%. The profile of the efficiency improvement during trigger the checking mechanism. It calculates the mismatch to
transient is presented in the Fig. 10. be less than 40; thus, the algorithm knows that the change is

1551-3203 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TII.2017.2703079, IEEE
Transactions on Industrial Informatics
7

due to uniform irradiance, not partial shading. It thus locates current (ΔI/I) is 0.03. On the other hand, under shading Pattern
the MPP at 0.8Voc_array (168 V). 1, current decreased from 2.262 to 2.261 A. This is based on
the respective voltages at 168 V and 172 V. Consequently,
ΔI/I = 0.00042. The data from TABLE IV clearly shows that
ΔI/I at STC is greater than ΔI/I at partial shading. Thus, the

Power (W)
Current (A)

condition dictated by [31] is correct. However, for Pattern 2,


the current is reduced from 3.58 to 3.22 A, at 168 V and 172
V, respectively. Since ΔI/I = 0.1, ΔI/I at STC is lower than
ΔI/I at partial shading. That is a clear violation of the proposed
condition in (14). Consequently, for this case, the method fails
Voltage (V)
to detect the occurrence of partial shading.
Fig. 13: Partial shading curve for the experiment
.
Uniform Irradiance Partial Shading Uniform Irradiance

Current (A)
VPV (V)

Tracking the MPP under ISTC

Power (W)
uniform G
Two point Tracking the global peak Two point
scanning
scanning
Partial shading
detected and whole
IPV (A)

PV curve Scanned

IPS
PPV (W)

Voltage (V)
(a)
Time (s)
ISTC
Fig. 14: Performance of [32] integrated with the proposed mechanism

Power (W)
under partial shading
Current (A)

B. Robustness Test
IPS
The robustness of the proposed scheme is compared with
work published in [31]. The latter is chosen because it also
utilizes the adaptive P&O. Unlike other general detection
mechanism (i.e. based on ΔP, ΔP/P or ΔI/I), it has the ability
to distinguish between uniform irradiance and partial shading. Voltage (V)
The working principal is based on measuring normalized (b)
current (ΔI/I) as follows. Fig.15: (a) I-V and P-V curve for pattern 1, (b) I-V and P-V curve for
 I   I   I 
     
pattern 2

 C U  C  STC  I C  PS
(14)
I I
where, Ic is the current at 0.8Voc_array (0.8NsVoc), while ΔI is the 171 V 171 V 167
VPV (V)

current change around the vicinity of 0.8Voc_array.


Despite the effectiveness of [38], it cannot guarantee the 82 V Shading
correct detection for every partial shading situation. To detection failed
illustrate this shortcoming, a specific case is presented here. Scanning for GP
Two shading patterns are generated for a 10×2 (S-P) system,
which result in the characteristic curves in Fig. 15. The I-V Time (s)
curve at STC and two particular partial shading condition is
(a)
shown in Figs. 15 (a) and (b), respectively. For such
configuration, Voc = 21.1 V, Voc_array = 210 V and Isc = 7.8 A. Partial shading occurrence checking
Two current points are marked of the curve at 0.8Voc_array (168
V) and another point as 172 V (due to ΔV=4 V, i.e. 2% of 171 V 171 V
VPV (V)

Voc_array)). Based on current values on these two voltages, 102 V


normalized current change (ΔI/I) is calculated in TABLE IV.
TABLE IV 82 V
CALCULATION OF ΔI/I ON SHADING PATTERN FROM FIG 15
STC condition Pattern 1 Pattern 2
I at 168 V 7.08 2.262 3.58
I at 172 V 6.86 2.261 3.22 Time (s)
(ΔI/I)STC 0.03 - - (b)
(ΔI/I)PS - 0.0004 0.1 Fig. 16: (a) Tracking profile [31] (b) Tracking profile for proposed method
At STC, when voltage moves from 168 to 172 V, the current The voltage (VPV) tracking for method [38] is presented in
decreases from 7.08 to 6.86 A. Thus, the normalized change in Fig. 16(a). When Pattern 2 is imposed (after 1 s), the

1551-3203 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TII.2017.2703079, IEEE
Transactions on Industrial Informatics
8

algorithm successfully detects the shading and initiate the 33 V and 90 V in Phases 3 and 4, respectively. As a result, a
search for the global peak. When the uniform irradiance is continuous varying global peak is created. In Phase 5, entire
imposed again (at 2 s), condition (14) is satisfied. The global array is shaded; thus a uniform irradiance is restored, albeit at
peak search is not initiated because it acknowledges the 200 W/m2. In subsequent phases, i.e. 6 and 7, two more
presence of uniform irradiance. However, when shading shading patterns are created; they shift the global peak to 88 V
Pattern 3 is imposed (after 3 s), the algorithm fails to detect and 65 V, respectively. Finally, in Phase 8, the cloud totally
the occurrence of partial shading due to the violation of moves away from the array and the 600 W/m2 uniform
condition (14). Thus, instead of searching for the global peak, irradiance is reestablished.
it traps in a local peak at 167 V. The tracking profile of the The voltage profile of the emulated passing cloud is shown
proposed detection scheme (integrated with adaptive P&O) is in Fig. 17. In Phase 2, 3 and 4, as the condition changes from
presented in the Fig. 16(b). It can be seen that, after the first uniform to partial shading (or from one shading pattern to
partial shading takes place (at 1 s), the checking mechanism is another), the checking mechanism is triggered. By scanning
immediately initiated and the global peak is correctly found at two points and calculating the value of G1 and G2, the
82 V. After 2 s, the partial shading condition is lifted and occurrence of partial shading is confirmed. Subsequently, the
uniform irradiance is restored. Sensing the change in power, global peak search tracks the maximum peak at 53, 33 and 90
the algorithm scans the two designated points and confirms V, respectively. In Phase 5, uniform irradiance is restored at
that the situation is a uniform irradiance change, not partial 200 W/m2. Here, it can be seen that the uniform irradiance
shading. Thus it updates Voc_array using (12) and directly goes condition is successfully identified and the unique peak is
the vicinity of 0.8Voc_array (where the peak is located). When detected at 84 V. Thus, the search on the entire P-V curve is
Pattern 3 is imposed, the scheme triggers the checking avoided. In Phase 6 and 7, the partial shading occurs again.
mechanism and detect the occurrence of partial shading. Then The checking mechanism confirms these and it successfully
the global peak search takes place. The peak is correctly locates the global peak at 88 V and 65 V. In Phase 8, when
detected at 102 V. uniform irradiance is restored, uniform irradiance is identified
This test shows that the proposed scheme guarantees the by and algorithm goes to track the MPP directly.
correct detection of partial shading, where technique in [31]
D. Integration with PSO
fails. Moreover, the open circuit voltage (Voc_array) is
continuously updated without any sensors, whereas [31] a The proposed scheme can also be integrated into a soft
temperature sensor is needed. Nevertheless, it has to be computing algorithm, for example the PSO. There are several
acknowledged that, [31] requires only two samples for the variation of PSO-based MPPT [10, 21, 29, 40], but in this
decision to be made, while the proposed technique requires work, the algorithm in [21] is chosen. All the tuning
three. However, this drawback is not a serious issue because parameters are kept as in the original version [21] and partial
of time requirement for one additional sample is negligible. shading check is done based on ΔP/P. Fig. 18 shows the
experimental performance of PSO when a sudden change in G
C. Dynamic Shading Profile takes place. As can be observed from the power oscillogram,
The dynamic shading is mainly caused by a passing cloud the algorithm initiates the global peak search because ΔP/P
phenomena. To emulate the moving cloud condition, a 5×4 S- has crossed the threshold. It generates the particles (samples)
P array are subjected eight different phases of shading, as on the P-V curve, performs the search and locates the peak
illustrated in Fig. 17. Initially, all modules are exposed to a after 12 sampling instants. However, as noted, such searching
uniform irradiance of 600 W/m2. When a fragment of the is unnecessary as there is no partial shading.
cloud moves upwards, some part of the array is covered with a
shade of 200 W/m2. As the cloud keeps on moving upwards,
different shading patterns are generated in Phase 3 through 8.
Interestingly, in Phase 2 the global peak is at 53 V; it moves to

Fig. 17: Dynamic shading condition due to passing cloud (Phase 1-phase 8)

1551-3203 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TII.2017.2703079, IEEE
Transactions on Industrial Informatics
9

VPV (V)
partial shading, the mismatch between G1 and G2 is
Transient
scanning on
approximately one order of magnitude higher than the case of
whole PV curve uniform change in irradiance. Thus, MPPT is able to take
conclusive decision to precisely determine the occurrence of
IPV (A)

partial shading. To perform this check, only 3 samples are


required; but by doing so, it avoids the MPPT from
Transient unnecessary scanning the entire P-V curve (which may take
PPV (W)

efficiency
dropped 10-25 samples). The scheme has been integrated with the
ηavg=50%
adaptive P&O and PSO MPPT, with an improvement of the
transient efficiency of 30-35%. It is expected that, similar
Time (s)
integration can be done with other conventional (IC, HC) and
Fig. 18: Performance of [21] in original PSO version metaheuristic (DE, ACO, FF, CS, GA) MPPT techniques. For
the dynamic test, the MPPT is subjected to moving cloud
VPV (V)

Two point conditions; the detection scheme has demonstrated its ability
scanning
to correctly determine the occurrence of partial shading
consistently. Besides, the proposed method offers the
IPV (A)

opportunity to update open circuit voltage using the


information obtained from these two scanned points. Thus,
Transient
efficiency
temperature and irradiance sensors are no longer needed,
PPV (W)

improved which leads to the reduction in implementation cost and


ηavg=85% complexity.
Time (s)
VI. REFERENCES
Fig. 19: Performance of [21] integrated with the proposed mechanism
Uniform Irradiance Partial Shading Uniform Irradiance
[1] J. Ahmed and Z. Salam, "A critical evaluation on maximum power
point tracking methods for partial shading in PV systems,"
VPV (V)

Tracking the MPP under Renewable and Sustainable Energy Reviews, vol. 47, pp. 933-953,
uniform G
Two point 2015.
scanning Tracking the global peak Two point
Partial shading scanning [2] Z. Lin, C. Yan, G. Ke, and J. Fangcheng, "New Approach for
detected and whole MPPT Control of Photovoltaic System With Mutative-Scale Dual-
IPV (A)

PV curve Scanned
Carrier Chaotic Search," Power Electronics, IEEE Transactions
on, vol. 26, pp. 1038-1048, 2011.
[3] L. Peng, L. Yaoyu, and J. E. Seem, "Sequential ESC-Based Global
MPPT Control for Photovoltaic Array With Variable Shading,"
PPV (W)

Sustainable Energy, IEEE Transactions on, vol. 2, pp. 348-358,


2011.
[4] K. Lian, J. Jhang, and I. Tian, "A maximum power point tracking
Time (s) method based on perturb-and-observe combined with particle
swarm optimization," Photovoltaics, IEEE Journal of, vol. 4, pp.
Fig. 20: Performance of [21] integrated with the proposed mechanism 626-633, 2014.
under partial shading [5] S. Mohanty, B. Subudhi, and P. K. Ray, "A new MPPT design
Fig. 19, shows the results when the proposed detection using grey wolf optimization technique for photovoltaic system
mechanism is incorporated with the original PSO. It can be under partial shading conditions," IEEE Transactions on
seen, the MPPT confirms the absence of partial shading Sustainable Energy, vol. 7, pp. 181-188, 2016.
[6] D. Teshome, C. H. Lee, Y. W. Lin, and K. L. Lian, "A Modified
because the mismatch is less than 40 W/m2. Accordingly, it Firefly Algorithm for Photovoltaic Maximum Power Point
searches for the MPP without scanning the entire P-V curve. Tracking Control Under Partial Shading," IEEE Journal of
Fig. 20 illustrates the performance of PSO under partial Emerging and Selected Topics in Power Electronics, vol. PP, pp.
shading when proposed mechanism is integrated. Since the 1-1, 2016.
[7] K. Sundareswaran, S. Peddapati, and S. Palani, "MPPT of PV
mismatch is higher than 40 W/m2, the algorithm recognizes systems under partial shaded conditions through a colony of
that partial shading has occurred and it initiates the PSO for flashing fireflies," Energy Conversion, IEEE Transactions on, vol.
global peak searching. It is presented in [22], that the most 29, pp. 463-472, 2014.
effective way to initiate three particles of PSO is 0.9Voc_array [8] R. Kotti and W. Shireen, "Efficient MPPT control for PV systems
adaptive to fast changing irradiation and partial shading
(near the end section of the P-V curve), 0.5Voc_array (middle of conditions," Solar Energy, vol. 114, pp. 397-407, 2015.
the P-V curve) and 0.8Voc (at the beginning of the P-V curve). [9] K. Sundareswaran, V. Vigneshkumar, P. Sankar, S. P. Simon, P. S.
Proposed mechanism calculate the Voc_array directly and R. Nayak, and S. Palani, "Development of an Improved P&amp;O
successfully re-initialize the particles. Afterwards it detects the Algorithm Assisted Through a Colony of Foraging Ants for MPPT
in PV System," IEEE Transactions on Industrial Informatics, vol.
global peak within 15 samples. 12, pp. 187-200, 2016.
[10] L. Yi-Hwa, H. Shyh-Ching, H. Jia-Wei, and L. Wen-Cheng, "A
V. CONCLUSION Particle Swarm Optimization-Based Maximum Power Point
Tracking Algorithm for PV Systems Operating Under Partially
In this paper, an accurate detection scheme to determine the Shaded Conditions," Energy Conversion, IEEE Transactions on,
occurrence of partial shading is proposed. The technique scans vol. 27, pp. 1027-1035, 2012.
two designated points on I-V curve and the recorded currents [11] K. Sundareswaran and S. Palani, "Application of a combined
particle swarm optimization and perturb and observe method for
are used to calculate irradiance levels, i.e. G1 and G2. During

1551-3203 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TII.2017.2703079, IEEE
Transactions on Industrial Informatics
10

MPPT in PV systems under partial shading conditions," Renewable System Under Partial Shading Condition," Industrial Electronics,
Energy, vol. 75, pp. 308-317, 2015. IEEE Transactions on, vol. 60, pp. 3195-3206, 2013.
[12] M. Seyedmahmoudian, R. Rahmani, S. Mekhilef, A. Maung Than [30] C. Kai, T. Shulin, C. Yuhua, and B. Libing, "An Improved MPPT
Oo, A. Stojcevski, S. Tey Kok, et al., "Simulation and Hardware Controller for Photovoltaic System Under Partial Shading
Implementation of New Maximum Power Point Tracking Condition," Sustainable Energy, IEEE Transactions on, vol. 5, pp.
Technique for Partially Shaded PV System Using Hybrid DEPSO 978-985, 2014.
Method," Sustainable Energy, IEEE Transactions on, vol. 6, pp. [31] A. Kouchaki, H. Iman-Eini, and B. Asaei, "A new maximum
850-862, 2015. power point tracking strategy for PV arrays under uniform and
[13] A. soufyane Benyoucef, A. Chouder, K. Kara, and S. Silvestre, non-uniform insolation conditions," Solar Energy, vol. 91, pp. 221-
"Artificial bee colony based algorithm for maximum power point 232, 2013.
tracking (MPPT) for PV systems operating under partial shaded [32] H. Patel and V. Agarwal, "Maximum Power Point Tracking
conditions," Applied Soft Computing, vol. 32, pp. 38-48, 2015. Scheme for PV Systems Operating Under Partially Shaded
[14] L. L. Jiang, D. L. Maskell, and J. C. Patra, "A novel ant colony Conditions," Industrial Electronics, IEEE Transactions on, vol. 55,
optimization-based maximum power point tracking for pp. 1689-1698, 2008.
photovoltaic systems under partially shaded conditions," Energy [33] K. Kobayashi, I. Takano, and Y. Sawada, "A study of a two stage
and Buildings, vol. 58, pp. 227-236, 3// 2013. maximum power point tracking control of a photovoltaic system
[15] L. Jiang and D. L. Maskell, "A simple hybrid MPPT technique for under partially shaded insolation conditions," Solar Energy
photovoltaic systems under rapidly changing partial shading Materials and Solar Cells, vol. 90, pp. 2975-2988, 2006.
conditions," in Photovoltaic Specialist Conference (PVSC), 2014 [34] F. Zhang, K. Thanapalan, A. Procter, S. Carr, and J. Maddy,
IEEE 40th, 2014, pp. 0782-0787. "Adaptive hybrid maximum power point tracking method for a
[16] R.-Y. Kim and J.-H. Kim, "An Improved Global Maximum Power photovoltaic system," Energy Conversion, IEEE Transactions on,
Point Tracking Scheme under Partial Shading Conditions," in vol. 28, pp. 353-360, 2013.
Journal of International Conference on Electrical Machines and [35] K. Ishaque and Z. Salam, "An improved modeling method to
Systems Vol, 2013, pp. 65-68. determine the model parameters of photovoltaic (PV) modules
[17] L. L. Jiang, D. Nayanasiri, D. L. Maskell, and D. Vilathgamuwa, using differential evolution (DE)," Solar Energy, vol. 85, pp. 2349-
"A hybrid maximum power point tracking for partially shaded 2359, 2011.
photovoltaic systems in the tropics," Renewable Energy, vol. 76, [36] M. G. Villalva and J. R. Gazoli, "Comprehensive approach to
pp. 53-65, 2015. modeling and simulation of photovoltaic arrays," Power
[18] J. Ahmed and Z. Salam, "An Improved Method to Predict the Electronics, IEEE Transactions on, vol. 24, pp. 1198-1208, 2009.
Position of Maximum Power Point During Partial Shading for PV [37] K. Ishaque, Z. Salam, H. Taheri, and Syafaruddin, "Modeling and
Arrays," Industrial Informatics, IEEE Transactions on, vol. 11, pp. simulation of photovoltaic (PV) system during partial shading
1378-1387, 2015. based on a two-diode model," Simulation Modelling Practice and
[19] S. A. Rizzo and G. Scelba, "ANN based MPPT method for rapidly Theory, vol. 19, pp. 1613-1626, 2011.
variable shading conditions," Applied Energy, vol. 145, pp. 124- [38] J. Ahmed and Z. Salam, "A Modified P&O Maximum Power Point
132, 2015. Tracking Method With Reduced Steady-State Oscillation and
[20] A. Ramyar, H. Iman-Eini, and S. Farhangi, "Global Maximum Improved Tracking Efficiency," IEEE Transactions on Sustainable
Power Point Tracking Method for Photovoltaic Arrays under Energy, vol. 7, pp. 1506-1515, 2016.
Partial Shading Conditions," IEEE Transactions on Industrial [39] F. Paz and M. Ordonez, "Zero oscillation and irradiance slope
Electronics, vol. PP, pp. 1-1, 2016. tracking for photovoltaic MPPT," Industrial Electronics, IEEE
[21] M. Miyatake, M. Veerachary, F. Toriumi, N. Fujii, and H. Ko, Transactions on, vol. 61, pp. 6138-6147, 2014.
"Maximum Power Point Tracking of Multiple Photovoltaic Arrays: [40] H. Renaudineau, F. Donatantonio, J. Fontchastagner, G. Petrone,
A PSO Approach," Aerospace and Electronic Systems, IEEE G. Spagnuolo, J. P. Martin, et al., "A PSO-Based Global MPPT
Transactions on, vol. 47, pp. 367-380, 2011. Technique for Distributed PV Power Generation," Industrial
[22] J. Ahmed and Z. Salam, "A Maximum Power Point Tracking Electronics, IEEE Transactions on, vol. 62, pp. 1047-1058, 2015.
(MPPT) for PV system using Cuckoo Search with partial shading
capability," Applied Energy, vol. 119, pp. 118-130, 2014.
[23] M. F. N. Tajuddin, S. M. Ayob, Z. Salam, and M. S. Saad, Jubaer Ahmed (M’2016) received the B.Sc.
"Evolutionary based maximum power point tracking technique degree in Electrical and Electronics Engineering
using differential evolution algorithm," Energy and Buildings, vol. from Bangladesh University of Engineering and
67, pp. 245-252, 2013. Technology (BUET) (2012) and Ph.D. from
[24] K. Punitha, D. Devaraj, and S. Sakthivel, "Artificial neural Universiti Teknologi Malaysia (UTM) (2016).
network based modified incremental conductance algorithm for Currently he is working as a lecturer in
maximum power point tracking in photovoltaic system under Swinburne University of Technology Sarawak,
partial shading conditions," Energy, vol. 62, pp. 330-340, 2013. Malaysia. His research interests include
[25] J. Young-Hyok, J. Doo-Yong, K. Jun-Gu, K. Jae-Hyung, L. Tae- photovoltaic modeling and control, energy
Won, and W. Chung-Yuen, "A Real Maximum Power Point conversion from renewable sources and power
Tracking Method for Mismatching Compensation in PV Array electronics.
Under Partially Shaded Conditions," Power Electronics, IEEE
Transactions on, vol. 26, pp. 1001-1009, 2011. Zainal Salam (M’1990) received the B.Sc degree
[26] H. Renaudineau, A. Houari, J. P. Martin, S. Pierfederici, F. from University of California, Chico (1985),
Meibody-Tabar, and B. Gerardin, "A new approach in tracking M.E.E. from Universiti Teknologi Malaysia
maximum power under partially shaded conditions with (UTM), Kuala Lumpur, Malaysia (1989) and Ph.D.
consideration of converter losses," Solar Energy, vol. 85, pp. 2580- from the University of Birmingham, U.K (1997).
2588, 2011. He has been a Lecturer at UTM for 30 years and is
[27] K. Sundareswaran, P. Sankar, P. S. R. Nayak, S. P. Simon, and S. now a Professor of Power Electronics and
Palani, "Enhanced Energy Output From a PV System Under Partial Renewable Energy at the Faculty of Electrical
Shaded Conditions Through Artificial Bee Colony," IEEE Engineering.
Transactions on Sustainable Energy, vol. 6, pp. 198-209, 2015.
Currently, he is the Director of the Inverter Quality Control Centre at
[28] N. Tat Luat and L. Kay-Soon, "A Global Maximum Power Point
UTM which is responsible for testing photovoltaic inverters that are to be
Tracking Scheme Employing DIRECT Search Algorithm for
connected to the local utility grid. Dr. Salam served the Editor of the IEEE
Photovoltaic Systems," Industrial Electronics, IEEE Transactions
TRANSACTION SUSTAINABLE ENERGY (2011-2013). His research interests
on, vol. 57, pp. 3456-3467, 2010.
include all areas of power electronics, renewable energy, power systems
[29] K. Ishaque and Z. Salam, "A Deterministic Particle Swarm
and machine control.
Optimization Maximum Power Point Tracker for Photovoltaic

1551-3203 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

You might also like