IET Power Electronics

Research Article

Improved hybrid algorithms-based MPPT ISSN 1755-4535

Received on 5th July 2018
Revised 13th November 2018
algorithm for PV system operating under Accepted on 4th December 2018
E-First on 25th January 2019
severe weather conditions doi: 10.1049/iet-pel.2018.5651
www.ietdl.org

Khaled Bataineh1
1Department of Mechanical Engineering, Jordan University of Science and Technology, Irbid, Jordan
E-mail: [email protected]

Abstract: This study aims to provide a comparison between several maximum power point tracking (MPPT) algorithms for PV
system under severe weather conditions. Two MPPT hybrid algorithms are proposed, namely (a) hybrid combination of fuzzy
logic controller (FLC) and the incremental conductance (IC) and (b) MPPT controller integrates FLC and perturb and observe
(P&O) method. The two developed hybrid algorithms combined the strength of both P&O, IC with FLC in single framework.
MATLAB/Simulink is used to investigate the response of both algorithms. Several weather conditions are simulated: (i) uniform
irradiation, (ii) sudden changing, and (iii) partial shading. Under partial shading on PV panel, multi-peaks appear in power–
voltage characteristics of the panel. Simulation results showed that ability of FLC to track MPP degrade significantly when
testing at weather conditions far from those used for training. Finally, the proposed hybrid algorithms successfully eliminate the
previous limitations associated with FLC, IC, and P&O algorithms individually with efficiency exceeds 97%.

Nomenclature points under partially shaded conditions [6]. Employing MPPT

techniques are considered the most economical ways of improving
Cmin minimum capacitance value, F the overall efficiency of the system compared with methods that
D duty cycle rely on the improving solar cell fabrication [7].
dV voltage difference, V Several methods for finding MPP are developed over the last
Fs switching frequency, Hz three decades [7–14]. These methods vary in terms of requiring
Im maximum current, A sensors, cost, efficiencies, complexity, correct tracking when
Iph photocurrent, A sudden shading or temperature changes, and in convergence speed.
Ipv photovoltaic current, A Esram and Chapman [8] presented a review of 19 methods for
Ir irradiation, W/m2 finding MPP. Generally, the techniques are classified into three
Lmin minimum inductance value, H classes: offline, online, and hybrid methods. Offline methods are
short-circuit current method, open-circuit voltage (OCV) method,
Pm maximum power, W look-up table method, curve-fitting based [9], and artificial
Pmax maximum power, W intelligence algorithms. Online methods are extremum seeking
Ppv photovoltaic Power, W control (ESC) method, ripple correlation control, hill climbing
R resistance, Ω (HC) [10], the incremental conductance (IC) method, perturbation
Rs series resistance, Ω and observation (P&O) method [11], and modified P&O [12, 13].
S step size As mentioned previously, when PV system is subjected to
T period time, s partially shaded condition, multiple peaks appeared in the P–V
Vm maximum voltage, V curve. Conventional methods such as P&O, HC, IC, direct search
Vo output voltage, V algorithm, and line search algorithm with Fibonacci sequence
Vpv photovoltaic voltage, V method could miss the global MPP [8, 14–16]. To overcome the
problem with the conventional method, several researchers have
Vs input voltage of converter, V
proposed several improvements. Due the fact that P&O and IC
Subscript L load methods use a fixed step size when tracking MPP, variable step-
size methods is developed to improve the steady-state performance
1 Introduction and dynamic response of PV system [17]. Tafticht et al. [18]
proposed a new method that combines non-linear expression that
Owing to global warming along with high prices of fossil fuel and
based on the OCV with P&O method to improve tracking
its hazards on the environment, searching for other renewable
efficiency of MPP. Recently, a new approach has been suggested
green sources of energy drew the attention of the world. Solar
by Heydari-doostabad et al. [19] based on extremum seeking
energy is considered the main source of renewable energy. Solar
control (ESC). It takes the advantage of a band-pass filter (PF)
energy is a permanent, none polluting, and low running cost source
using high PF, and low PF by only passing input power frequencies
of energy. Photovoltaic (PV) (solar cell) systems are one of the
that include derivative of PV with respect to its voltage, and so the
most favourable systems and their installation is spreading widely.
system will operate at the global maximum point. Noguchi et al.
PV systems can be connected to the grid or can be used as stand-
[20] proposed a short-circuit pulse-based MPPT with a fast scan on
alone systems [1, 2]. The power generated from PV panel depends
the P–V curve to identify the proportional parameter which is
on the amount of solar irradiance, cell temperature, and load [3–5].
commonly used in a current-based MPPT [21]. Although the
Maximum power point tracking (MPPT) is essential to keep the
proposed method successfully found the global maximum point,
system operating at its optimal power. Up to date, the overall PV
however, momentary power loss is accompanied with additional
efficiency reaches around 15%. Raising power generated from PV
extra cost. To avoid this extra power loss, Kazmi et al. [22]
systems can be achieved by tracking the MPP of the output power–
proposed a controller that fluctuates the converter's duty cycle from
voltage (P–V) curve. This curve may contain multi-local maximum
zero to one to measure the OCV and the short-circuit current and

IET Power Electron., 2019, Vol. 12 Iss. 4, pp. 703-711 703

© The Institution of Engineering and Technology 2018
 17554543, 2019, 4, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-pel.2018.5651, Wiley Online Library on [19/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
technique for a PV system based on the integration between
Lagrange interpolation (LI) and PSO for eliminating the traditional
methods problems [31]. Using LI allows the particles to migrate
very close to the best position directly and hence avoids local
minima trapping. Such technique minimises the PSO number of
iterations for reaching the MPP. The results are compared with
perturb and observe (P&O), IC, and the conventional PSO. The
comparison indicated that the proposed technique enhances the
search stability and the tracking speed capability under any abrupt
change. Daraban et al. [32] proposed an MPPT technique that
embeds P&O algorithm in genetic algorithms structure. Each
individual carrying an information of the reference voltage, the
direction, and the step value. Lian et al. presented a hybrid MPPT
algorithm that combines P&O and PSO methods [34]. Initially,
Fig. 1 Effect of irradiance changes on I–V characteristics P&O method is employed to locate the nearest local maximum.
Then, the PSO method is employed to locate the MPP starting from
that point. Sundareswaran et al. [35] developed a new algorithm
for MPPT in large PV systems under partial shading conditions.
PSO is used for MPPT during the initial stages of tracking and then
employs the traditional P&O method at the final stages. Bataineh
and Taamneh proposed an MPPT algorithm based on a hybrid
combination between the FLC and the conventional P&O method
[36]. The FLC is used to locate the region of MPP, then, P&O
algorithm is utilised to accurately track the MPP.
The main objective of this paper is to develop an efficient
MPPT hybrid algorithms that overcome the problems with
conventional methods under sudden changes and partial shading
conditions. The proposed algorithm is a hybrid combination
between the FLC and the IC. A further objective is to compare
between the performance of two hybrid algorithms between FLC
and P&O [36] and the current proposed hybrid algorithm under
several weather conditions. Utilising FLC rapidly brings the system
Fig. 2 Characteristics curves of PV under partial shading to the vicinity of the MPP and allows the use of a small step size in
the P&O and IC algorithm for higher accuracy and lower
then computes the optimum voltage and current. On the basis of oscillations. The performance of the proposed algorithms is
the computed values, in single step, the operating point is moved to investigated by building MATLAB/Simulink models consisting of
the optimal operating point. The conventional HC algorithm is the PV system, boost converter, and controllers. Fuzzy controller's
utilised to maintain the system operating around the maximum parameters and membership functions are optimised employing the
point. Although the system finds the global maximum, however, adaptive neuro-fuzzy inference system (ANFIS). Comparisons
the significant loss in power is experienced. Utilising practical against stand-alone FLC, IC, and P&O controllers are performed to
swarm optimisation algorithm under abnormal weather conditions assess the performance of the proposed hybrid algorithm.
results in a long computation time to reach the maximum operating
point [23]. 2 PV modelling and characteristics
Intelligent systems such as neural networks (NNs) and fuzzy The detailed description of PV modelling can be found in [2].
logic controllers (FLCs) have been used successfully in tracking Current–voltage (I–V) non-linear characteristic curves are shown in
MPP of PV to decrease computation power requirement, while Fig. 1. Partial shading occurs when radiation is not equally
increasing speed and efficiency of the tracking [2]. They are robust distributed on PV cells. The current generated by shaded cells is
and relatively simple to design. However, they require complete decreased. This leads to a reduction in the overall power generated
knowledge of the operation of the PV system by the designer. from the PV system. To understand such phenomena, a PV array
Othmana et al. [24] validated the ability of FLC to find MPP system with modules connected in series is considered. Under
comparing with P&O algorithm. Punithaa et al. [25] used a partial shading condition, multiple peaks are presented in the P–V
modified IC method with NN to supply reference voltage (Vref) and characteristic (see Fig. 2).
compared the results with FLC and P&O approaches to validate
their proposed method. Results showed the highest performance
with the least response time when using artificial NN (ANN) with 3 MPPT algorithms
IC. Subiyanto et al. [26] presented a new method using Hopfield In this section, a brief description about the P&O, IC, and FLC is
NN to tune FLC parameters to enhance robustness and accuracy. given.
Several hybrid MPPT algorithms have been proposed by many
researchers recently [27–36] to deal with partial shading 3.1 P&O controller
conditions. A hybrid controller that combines the advantages of
fuzzy logic and conventional proportional–integral–derivative The detailed description of P&O controller can be found in [37–
control is proposed by Wang et al. [27] and Dounis et al. [28]. 39]. The basic idea of this controller is to provoke perturbation by
Jiang et al. [29] proposed a hybrid MPPT controller that combines acting (decrease or increase) on the pulse-width modulation duty
ANN and P&O without irradiance sensor. ANN, which is used to cycle and observing the effect on the output PV power. To generate
train several offline partial shading cases, is used to predict the maximum power, DC–DC boost converter is used and placed
region of MPP; then, P&O technique is employed to accurately between the source and the load. To simulate P&O algorithm, PV
track the MPP. Seyedmahmoudian et al. [30] proposed a hybrid system composed of PV panel, boost DC–DC converter, MPPT,
evolutionary algorithm called the DEPSO technique, a combination and the resistive load is built as shown in Fig. 3. In this work, a
of the differential evolutionary algorithm and particle swarm boost converter controlled by MPPT algorithm is used to track
optimisation (PSO) [33]. Their simulation results proved the MPP. The voltage gain of the converter is given as [40]
advantages of the proposed method such as its reliability, system
independence, and accuracy in tracking the global MPP under
partial shading conditions. Ramdan et al. proposed a new MPPT
704 IET Power Electron., 2019, Vol. 12 Iss. 4, pp. 703-711
© The Institution of Engineering and Technology 2018
 17554543, 2019, 4, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-pel.2018.5651, Wiley Online Library on [19/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
VO 1 3.3 Adaptive neuro-fuzzy inference system
= (1)
VS 1−D
The fuzzy-based controllers can effectively deal with non-
linearities and mimic the human expertise to form an
The minimum values of inductance and capacitance of the approximation that maps input values to their predicted outputs
converter necessary for stability are given as [40] (listed in Table 1) based on IF–THEN rules. The fuzzy inference is carried out using
2
Sugeno-type system. The Sugeno method is chosen in this work
DR 1 − D because it is computationally efficient, works well with linear
Lmin = (2)
2FS techniques, works well with optimisation and adaptive techniques,
and it is well suited to mathematical analysis. The defuzzification
V OD uses the centre of gravity to compute the output of this FLC, which
Cmin = (3)
FSΔV OR is the duty cycle
n
P&O algorithm has been implemented in the Simulink model to ∑ j = 1 μ(D j) − D j
control the duty cycle of the switching signal of the converter. ΔV D= n (5)
∑ j = 1 μ(D j)
and ΔP are used to detect irradiation variations, and the algorithm
determines the value of the duty cycle necessary to attain MPP on
the load. It changes the duty cycle value by a step size (S) which is Fuzzy controller's parameters and membership functions developed
determined by the designer. in this paper are optimised by employing the ANFIS. A training set
of 55 data points are used. ANFIS is a single framework that
combines both concepts of ANN fuzzy logic and ANN. The
3.2 IC algorithm ANFIS generates rules and membership function parameters and
IC was designed based on an observation of P–V characteristic tune them based on a given input–output data set. The inputs of the
curve. This algorithm was developed in 1993 and was intended to controller are the OCV (Voc) and the short-circuit current (Isc), and
overcome some drawbacks of P&O algorithm. IC aims to improve the output is the desired duty cycle (D). To train the NN and tune
the tracking time and to generate more energy under fast changes the fuzzy controller parameters, a set of input–output data is
of weather conditions. The MPP can be calculated by using the obtained manually using MATLAB/Simulink models. The
relation between dI/dV and –I/V. If dP/dV is negative, then MPPT parameters used to train the FLC using ANFIS are listed in Table 2.
is lying on the right-hand side of the recent position and if the MPP
is positive the MPPT is on the left-hand side. The equation of IC The NNs algorithm is operated based on internal data training,
method is while external data training is used for fuzzy logic algorithm
operation. The tracking error and change in error are fed as input to
dP dVI dV dI dI NN while the NN output is fed as an input to the fuzzy system. The
= =I +V = I+V (4)
dV dV dV dV dV OCV is amplified to increase system's sensitivity to overcome the
narrow range at which it changes as shown in Fig. 5a. Each input
The flowchart of the IC algorithm is shown in Fig. 4. comprises five Gaussian-type membership functions as shown in
Fig. 5b. The 25 rules of the FLC are shown in Fig. 6. To train the
NN and tune the fuzzy controller parameters, a set of input–output
data was obtained manually using a MATLAB/Simulink models.
Two models were built; the first model was used to record the
maximum available power at each insolation level ranging from
100 to 1000 W/m2 in partial and full shading conditions. The
second model was used to seek the optimum duty cycle manually
in an iterative manner and to record the corresponding OCV and
short-circuit current. Using the grid partitioning algorithm and a
hybrid optimisation method, 3000 epochs are trained to produce

Table 1 DC–DC booster converter design values

Electrical characteristic Values
inductance 9 × 10−4 H
output capacitor 0.001 F
input capacitor 1 × 10−9 F
Fig. 3 Schematic diagram of PV system with MPPT resistance load 28.18 Ω

the fuzzy inference system. The structure of ANFIS is organised

into two parts: the introductory and concluding, which are attached

Table 2 ANFIS-editor training parameters

Fuzzy logic type Sugeno
number of inputs 2
number of membership function 10
number of training epochs 3000
input membership function type Gaussian
output membership function type linear
algorithm used grid partitioning
root-mean-square error considered 0.034
validation set 55
Fig. 4 Flowchart of the IC algorithm optimisation method hybrid

IET Power Electron., 2019, Vol. 12 Iss. 4, pp. 703-711 705

© The Institution of Engineering and Technology 2018
 17554543, 2019, 4, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-pel.2018.5651, Wiley Online Library on [19/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 7 ANFIS structure

3.5 IC and FLC hybrid MPPT algorithm

The idea of the proposed algorithm is inspired by using the benefits
of both FLC and IC algorithm in a single framework. As
mentioned previously, the FLC has the capability to work under
severe weather conditions with low accuracy, while IC can be
highly accurate when using small step size. An efficient algorithm
Fig. 5 Fuzzy logic controller
can be constructed when utilising the FLC to drive the system
(a) Structure of FLC, (b) FLC's input membership functions
closer to the MPP, then using IC to accurately locate the global
MPP. Hence, the proposed algorithm merges the swiftness of
approximation of the fuzzy system with the accuracy of IC method.

4 Simulation results and discussion

To compare and evaluate the performance of proposed hybrid
algorithms, a comparison between their performance with classical
P&O method alone, IC controller alone, and FLC alone are
presented. The performance of the proposed FLC is evaluated
using MATLAB/Simulink. The simulated model consists of 72
cells connected in series to form PV module. Partially shaded
model is simulated by connecting two modules in series allowing
each module receives a different level of irradiation. The five
algorithms were simulated under the following tests: (i) uniform
irradiation, (ii) sudden changing, and (iii) partial shading.

4.1 Performance of P&O, IC, and FLC controller alone

The electrical specifications of PV cell are: Isc = 4.75 A, Voc = 0.6
V, and Rs = 5.1 × 10−3 Ω. The values of power listed in Table 4
demonstrate the importance of using MPPT. Table 5 shows that
Fig. 6 Rules of the FLC
under uniform radiations, the three studied algorithms successfully
reach the MPP. However, under partial shading conditions, both
by a set of rules. ANFIS network is a multi-layer network that has
P&O and IC got trapped in the local MPP, where FLC has
five distinct layers and all nodes in a given layer have the same
approached global MPP. It is worth mentioning that FLC requires a
function. Fig. 7 shows the architecture of ANFIS network with two
large number of training data points to work well. It has been
inputs and one output of a first-order Sugeno-type fuzzy system.
reported that fuzzy controller accuracy decreases significantly
The training data points that are used to train the FLC are
when tested far from the training data points. It is worth
shown in Table 3. Table 3 lists OCV, the short-circuit current, the
mentioning that using large number of data for training is not
optimal duty cycle, output power, and the efficiency. The tracking
practical to enhance the accuracy of FLC, because this required
efficiency is defined as [2]
large number of membership functions that consume higher
processing time. To clarify this point, we simulated partial shading
Ppv
efficiency = × 100% (6) cases where the amount of solar radiation is far from those
pmpp implemented to train the fuzzy controller. Results presented in
Table 6 demonstrate that the accuracy of FLC has dropped.
Fig. 9 shows the output power for fuzzy controller compared
3.4 P&O and FLC hybrid MPPT algorithm with P&O algorithm for several partial shading conditions. The
irradiation applied on the simulated model at this case is equally on
The main motivation of combining the FLC and P&O algorithms is
the whole panel, and suddenly decreased partially at 0.2 s. Fuzzy
to take the advantages of the strength of both FLC and P&O
controller found the peak at the first part with accuracy more than
algorithm in a single framework. FLC can deal with the dynamic
P&O algorithm. After partial shading being applied on 36 cells of
changes of weather conditions in the price of low accuracy, while
the PV panel, fuzzy controller tracked the global MPP while the
P&O algorithm can achieve high accuracy when using small step
P&O algorithm trapped at the local. Fuzzy controller has reached
size. An efficient and robust algorithm can be developed by using
the global point with an efficiency of 99% while P&O controller
FLC to locate the region of MPP and then utilising the P&O to
failed to reach this point. Fuzzy controller detects irradiation
accurately and rapidly find MPP. Hence, the proposed hybrid
changing by detecting changes in voltage and power values of PV
algorithm enhances the search stability and the tracking speed
system. This controller has the ability to change the duty cycle by a
capability. Schematic diagram of PV system with hybrid MPPT
precise difference depending on membership functions. P&O
used to build MATLAB\Simulink model is shown in Fig. 8.
changes the value of duty cycle by a fixed step every time
irradiation changes. Owing to the fixed step value, it does not reach
706 IET Power Electron., 2019, Vol. 12 Iss. 4, pp. 703-711
© The Institution of Engineering and Technology 2018
 17554543, 2019, 4, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-pel.2018.5651, Wiley Online Library on [19/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 3 Data used to train the FLC, inputs: OCV and the short-circuit current, output: optimal duty cycle
Ir1, W/m2 Ir2, W/m2 Voc, V Isc, I D Output power Efficiency
1000 1000 8.2 55.0 0.571 260.1 96.62
1000 900 8.2 53.73 0.556 241.6 96.48
1000 800 8.2 52.32 0.534 212.7 94.06
1000 700 8.2 50.71 0.496 191.8 96.16
1000 600 8.2 48.86 0.445 165.1 96.22
1000 500 8.2 46.67 0.4 136.8 95.64
1000 400 8.2 43.98 0.323 110 91.17
1000 300 8.2 40.52 0.676 116.8 98.29
1000 200 8.2 35.65 0.675 116.8 99.79
1000 100 8.2 27.31 0.669 115.1 99.9
900 900 7.38 52.47 0.546 232.4 96.79
900 800 7.38 51.05 0.522 212.6 96.17
900 700 7.38 49.44 0.496 189.5 96.43
900 600 7.38 47.59 0.447 164.1 96.69
900 500 7.38 45.4 0.397 136.8 96.51
900 400 7.38 42.72 0.323 109.1 96.35
900 300 7.38 39.26 0.685 107.1 99.75
900 200 7.38 34.38 0.701 99.46 94.66
900 100 7.38 26.05 0.715 99.3 95.47
800 800 6.56 49.63 0.521 203.9 96.52
800 700 6.56 48.03 0.495 185.5 96.62
800 600 6.56 46.17 0.448 160.5 96.02
800 500 6.56 43.98 0.398 134.8 96.24
800 400 6.56 41.3 0.327 107.6 95.82
800 300 6.56 37.84 0.658 94.71 98.96
800 200 6.56 32.96 0.642 92.1 97.71
800 100 6.56 24.63 0.644 91.8 98.92
700 700 5.74 46.42 0.496 175.8 96.22
700 600 5.74 44.57 0.451 157.1 96.29
700 500 5.74 42.38 0.399 132.8 96.2
700 400 5.74 39.69 0.327 107.1 96.51
700 300 5.74 36.23 0.637 81.8 97.56
700 200 5.74 31.36 0.633 80.5 97.52
700 100 5.74 23.02 0.628 81.1 99.75
600 600 4.92 42.72 0.449 148.1 95.89
600 500 4.92 40.52 0.402 129.3 96.06
600 400 4.92 37.84 0.345 105.1 96.25
600 300 4.92 34.38 0.247 79.09 96.45
600 200 4.92 29.51 0.618 69.25 97.92
600 100 4.92 21.17 0.605 69 99.11
500 500 4.1 38.33 0.419 121.6 96.06
500 400 4.1 35.65 0.32 102.1 95.99
500 300 4.1 32.19 0.226 77.8 96.38
500 200 4.1 27.31 0.615 54.2 92.32
500 100 4.1 18.98 0.605 54.6 94.46
400 400 3.28 32.96 0.32 95 95.78
400 300 3.28 29.51 0.215 75.1 95.64
400 200 3.28 24.63 0.045 50.8 96.45
400 100 3.28 16.3 0.565 43.6 95.05
300 300 2.46 26.05 0.23 69.72 96.36
300 200 2.46 21.17 0.065 49.3 96.36
300 100 2.46 12.84 0.515 31.11 91.77
200 200 1.64 16.3 0.07 44.8 96.78
200 100 1.64 7.961 0.418 20.1 82.96

the necessary precise value of the duty cycle to get maximum 4.2 Performance of the hybrid FLC and P&O algorithms
power from the system and might get trapped at local maximum
under partial shading conditions due to its way of tracking by As discussed before, FLC algorithms fail to find MPP when
looking for the first peak in power values. irradiations level far from those used to train its membership
functions. Therefore, a hybrid algorithm is proposed to use the
strength of both P&O and FLC in single framework in order to
track the MPP efficiently and accurately. Several weather

IET Power Electron., 2019, Vol. 12 Iss. 4, pp. 703-711 707

© The Institution of Engineering and Technology 2018
 17554543, 2019, 4, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-pel.2018.5651, Wiley Online Library on [19/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
conditions, namely (a) uniform condition, (b) partial shading, and show that the proposed algorithm rapidly finds the new MPP. For
(c) sudden changes are simulated to test the performance of the example, Fig. 10c shows the performance when the radiation
proposed hybrid algorithm. The performance of the hybrid changes from 900 to 250 W/m2, the power obtained from P&O
algorithm is compared against FLC alone and P&O alone. Under controller changed from 135 to 35 W in 0.4 s while it took only
uniform conditions, all algorithms successfully approach MPP (see 0.15 s for the proposed controller to change the power from 134 to
Table 5). Moreover, the hybrid algorithm has faster response and 35 W. It is clear from simulation results that the proposed hybrid
has the ability to deliver more power than P&O. P&O can deliver controller successfully outcomes the shortages of fuzzy and P&O
higher power if decreasing the step size which unavoidably algorithm alone.
accompanied by a slower response (see Fig. 10a). Table 7 compares between the performances of the proposed
To further evaluate the performance of the proposed algorithm, hybrid algorithm with FLC alone for several cases at weather
several partial shading cases are simulated. For all simulation cases conditions not utilised for training. It is clear that the hybrid
studied, the hybrid controller successfully finds the global MPP. algorithm was successfully able to harvest over 99% of the
Under partial shading condition, P&O got trapped in the local MPP maximum power available for all random weather conditions,
(see Fig. 10b). As mentioned earlier, the fuzzy controller always while the FLC efficiency decreases significantly when conditions’
finds the duty cycle that brings the system to the vicinity of the points are far from those used for training.
global MPP. The P&O part of the hybrid controller uses this duty
cycle as an initial guess to further increase the power obtained by
the PV module.
P&O algorithm has a relatively long recovery time when
exposed to sudden changes in radiation levels. To assess the
proposed algorithm under this condition, several simulations are
conducted by replacing the constant irradiation input of the PV
model with a step signal. The results showed in Figs. 7d and 10c

Table 4 Comparison between output power values with and

without MPPT
Case Irradiance Output Output Percentage
number level, W/m2 power, W power, W increase
(without (with
MPPT) MPPT)
1 800 102.5 118 18.7
2 600 64.9 87 34
3 500 45 70.8 57
4 300 21.4 38.7 80 Fig. 8 Schematic diagram of PV system with hybrid MPPT

Table 5 Performances of P&O, IC, and FLC alone under different weather conditions
Case number Ir1, W/m2 Ir2, W/m2 Nominal power, W P&O, W IC, W Power after fuzzy, W Ppv
Efficiency × 100%
pmpp
Uniform irradiation
1 1000 1000 149 149 149 149 100
2 900 900 133 133 133 133 100
3 800 800 118 117.5 117.5 117.5 99.6
4 700 700 102.5 101.5 101.5 101.5 99
5 600 600 87.2 87 87 87 99.8
6 500 500 71.3 70.8 70.8 70.8 99.3
7 400 400 56.5 56 56 56 99
8 300 300 41.5 41 41 41 99
9 200 200 26.6 24.5 24.5 24.5 92
Partial shading
10 1000 800 126.3 71.24 73.4 126 99.8
11 900 600 95.7 76.4 77.3 94 98
12 800 500 79.5 55.7 55.3 79 99.4
13 700 500 78.2 49.7 50.3 78 99.7
14 600 400 62.3 41.5 41.7 62 99.5
15 900 300 61.4 52.8 53.1 60 97.7
16 700 300 48 39.4 39.7 47.5 99
17 500 300 46.3 34.9 35.2 46 99.4
Values of model outputs.

Table 6 Performance of FLC alone under partial shading conditions at random points far from training data point
Case number Ir1, W/m2 Ir2, W/m2 Nominal power, W Power after fuzzy, W Efficiency, %
1 892 407 81.2 72.6 89
2 644 596 89.2 81.4 91
3 985 317 74.3 62.8 85
4 593 396 60.1 51.1 85

708 IET Power Electron., 2019, Vol. 12 Iss. 4, pp. 703-711

© The Institution of Engineering and Technology 2018
 17554543, 2019, 4, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-pel.2018.5651, Wiley Online Library on [19/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License

Fig. 9 Comparison between fuzzy and P&O partial shading Fig. 10 Performance of hybrid algorithm under
(a) Uniform (1000, 1000) to partial shading (1000, 800) W/m2, (b) Uniform (800, (a) Uniform condition 650 W/m2, (b) Partial shading conditions (900, 600) W/m2,
800) to partial shading (800, 500) W/m2, (c) Uniform (600, 600) to partial shading sudden changes in radiation levels, (c) Dropping from 900 to 250 W/m2, (d) Rising
(600, 400) W/m2, (d) Uniform (600, 600) to partial shading (500, 300) W/m2 from 350 to 750 W/m2

IET Power Electron., 2019, Vol. 12 Iss. 4, pp. 703-711 709

© The Institution of Engineering and Technology 2018
 17554543, 2019, 4, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-pel.2018.5651, Wiley Online Library on [19/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 7 Performance of hybrid algorithm under partial shading conditions at points not utilised in training
Case number Ir1, W/m2 Ir2, W/m2 Nominal power, W FLC alone, W Hybrid, W
1 892 407 81.2 72.6 79.1
2 644 596 89.6 81.5 87.3
3 400 200 35.6 28.7 34.8

Table 8 Comparison between the performance of hybrid and IC under uniform conditions
Case number I, W/m2 Nominal power, W IC (step size 0.001) power, IC (step size 0.00001) Hybrid, W power, W/
W/response time, s power, W/response time, s response time, s
1 1000 149 149/0.15 149/4.6 149/0.2
2 900 133 133/0.14 132.8/4.5 132.7/.25
3 800 118 116.7/0.17 117.5/3.7 117.0/0.22
4 700 102.5 100.4/0.15 102.3/3.7 101.7/0.21
5 uniform conditions 600 87.2 86.1/0.17 87.1/4.1 86.6/0.20
6 sudden changes 1000Ú800 118 116.3/0.26 117.1/3.5 116.8/0.28
7 700Ú600 87.2 86.3/0.29 87.2/3.7 87.0/0.27
8 500Ú300 41.5 39.6/0.27 40.7/4.2 40.5/0.25

Table 9 Performance of hybrid algorithm under partial shading conditions at points not utilised in training
Case number Ir1, W/m2 Ir2, W/m2 Nominal power, W IC alone, W Hybrid P&O and FLC Hybrid IC and FLC
Power, W/efficiency, % Power, W/efficiency, %
1 900 600 95.7 77.3 95.1 99 95.3 100
2 800 500 79.5 55.3 79.1 99 79.4 100
3 700 500 78.2 49.7 77.8 99 77.8 99
4 600 400 62.3 41.5 61.3 98 61.3 98
5 892 407 86.2 63.2 79.1 97 79.5 98
6 644 596 103.7 68.9 87.3 97 87.9 98
7 400 100 46.3 26.7 28.3 95 28.9 97

4.3 Performance of the hybrid FLC and IC algorithms performances of these algorithms under several weather conditions
are evaluated. On the basis of simulations results, it can be
In this section, the performance of a hybrid algorithm consisting of concluded that under:
FLC and IC algorithms is evaluated under several weather
conditions. The results presented in Table 8 shows that under • Uniform weather conditions: All MPPT algorithms successfully
uniform conditions, all algorithms successfully reach MPP. The reached MPP.
accuracy of IC increased when using related smaller time step in
• Sudden changes of weather conditions: All studied algorithms
the price of response time. However, the hybrid algorithm shows
showed the ability to recover from sudden changes and maintain
superior performance compared with other algorithms. Under
stability under partial shading condition FLC and hybrid
sudden changes of weather conditions, IC algorithm has a
algorithms have a better response than IC and P&O algorithms.
relatively long recovery time compassed hybrid algorithm.
The results listed in Table 8 show that the proposed algorithm • Under partial shading conditions: Both P&O and IC failed to
rapidly finds the new MPP. For example, the performance when the find MPP. FLC found MPP if weather conditions are closer the
data point used from training; otherwise, the accuracy has been
radiation changes from 1000 to 800 W/m2, the power obtained
compromised.
from IC controller changed from 149 to 117.1 W in 3.5 s while it
took only 0.28 s for the proposed hybrid controller to change the • The two developed hybrid algorithms effectively and accurately
power from 149 to 116.8 W. It is clear from simulation results that found global MPP under all weather conditions with efficiency
the proposed hybrid controller successfully outcomes the shortages exceeding 97%.
of fuzzy and IC algorithm alone. • Hybrid algorithms consist of IC and FLC slightly outperforms
Table 9 compares the performances of the proposed hybrid P&O and FLC algorithms for tracking global MPP under
algorithms under random weather conditions. It is clear that the random weather conditions.
hybrid algorithm was successfully able to harvest over 98% of the
maximum power available for all random weather conditions, 6 Acknowledgment
while the FLC efficiency decreases significantly when conditions’
This work was supported by the Deanship of Research of Jordan
points are far from those used for training. Finally, simulation
University of Science and Technology [grant number 476-2017].
results listed in Table 9 show that hybrid (IC and FLC) has slightly
better performance compared with P&O with FLC.
7 References
5 Conclusion [1] Bataineh, K.M., Dalalah, D.: ‘Optimal configuration for design of stand-alone
PV system’, Smart Grid Renew. Energy, 2012, 3, pp. 139–147
In this paper, a comparison between the performances of P&O, IC, [2] Allataifeh, A., Bataineh, K., Al-khedher, M.: ‘Maximum power point tracking
FLC, and two proposed hybrid algorithms for the MPPT of a PV using fuzzy logic controller under partial conditions’, Smart Grid Renew.
Energy, 2015, 6, pp. 1–13
system under variable insolation conditions have been presented. [3] Lu, C.F., Liu, C.C., Wu, C.J.: ‘Dynamic modeling of battery energy storage
The two developed hybrid algorithms combined the strength of system and application to power system stability’, IEE Proc. Gener. Transm.
both P&O and IC with FLC in a single framework. MATLAB/ Distrib., 1995, 142, (4), pp. 429–435
Simulink model consists of PV panel, a boost converter with [4] Möller, H.J.: ‘Semiconductors for solar cells’ (Artech House Inc., Norwood,
MA, 1993)
MPPT algorithms connected to a resistive load has been built in
order to evaluate the performance of the proposed controllers. The
710 IET Power Electron., 2019, Vol. 12 Iss. 4, pp. 703-711
© The Institution of Engineering and Technology 2018
 17554543, 2019, 4, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-pel.2018.5651, Wiley Online Library on [19/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
[5] Gottschalg, R., Rommel, M., Ineld, D.G., et al.: ‘Comparison of different [24] Othmana, A., El-arinia, M., Ghitasb, A., et al.: ‘Real world maximum power
methods for the parameter determination of the solar cell's double exponential point tracking simulation of PV system based on fuzzy logic control’, NRIAG
equation’. 14th European Photovoltaic Science and Engineering Conf. J. Astron. Geophys., 2012, 1, pp. 186–194
(PVSEC), Barcelona, Spain, 1997 [25] Punithaa, K., Devaraja, D., Sakthivel, S.: ‘Artificial neural network based
[6] Cuia, Y., Yaoa, W., Luoa, J.: ‘A research and improvement on a maximum modified incremental conductance algorithm for maximum power point
power point tracking method for PV system under partially shaded tracking in photovoltaic system under partial shading conditions’, Energy,
conditions’, Procedia Eng., 2012, 29, pp. 2583–2589 2013, 62, pp. 330–340
[7] Salam, Z., Ahmed, J., Merugu, B.: ‘The application of soft computing [26] Subiyanto, S., Mohamed, A., Hannan, M.: ‘Intelligent maximum power point
methods for MPPT of PV system: a technological and status review’, Appl. tracking for PV system using Hopfield neural network optimized fuzzy logic
Energy, 2013, 107, pp. 135–148 controller’, Energy Build., 2012, 51, pp. 29–38
[8] Esram, T., Chapman, P.L.: ‘Comparison of photovoltaic array maximum point [27] Wang, Y., Ding, L., Li, N.: ‘The application of fuzzy parameters self-tuning
tracking techniques’, IEEE Trans. Energy Convers., 2007, 22, pp. 439–449 PID controller in MPPT of photovoltaic power system’. Proc. Transportation,
[9] Bhatnagar, P., Nema, R.: ‘Comparison of photovoltaic array maximum point Mechanical, and Electrical Engineering (TMEE), Changchun, China, 2011,
tracking techniques maximum power point control techniques: state-of-the-art pp. 1129–1132
in photovoltaic applications’, Renew. Sustain. Energy Rev., 2013, 23, pp. 224– [28] Dounis, A.I., Kofinas, P., Alafodimos, C., et al.: ‘Adaptive fuzzy gain
241 scheduling PID controller for maximum power point tracking of photovoltaic
[10] Chia-Hung, L., Cong-Hui, H., Yi-Chun, D., et al.: ‘Maximum photovoltaic system’, Renew. Energy, 2013, 60, pp. 202–214
power tracking for the PV array using the fractional-order incremental [29] Jiang, L.L., Nayanasiri, D., Maskell, D.L., et al.: ‘A hybrid maximum power
conductance method’, Appl. Energy, 2011, 88, pp. 4840–4847 point tracking for partially shaded photovoltaic systems in the tropics’,
[11] RezaReisi, A., Moradi, M., Jamasb, S.: ‘Classification and comparison of Renew. Energy, 2015, 76, pp. 53–65
maximum power point tracking techniques for photovoltaic system: a review’, [30] Seyedmahmoudian, M., Rahmani, R., Mekhilef, S., et al.: ‘Simulation and
Renew. Sustain. Energy Rev., 2013, 19, pp. 433–443 hardware implementation of new maximum power point tracking technique
[12] Mamarelis, E., Petrone, G., Spagnuolo, G.: ‘A two-steps algorithm improving for partially shaded PV system using hybrid DEPSO method’, IEEE Trans.
the P&O steady state MPPT efficiency’, Appl. Energy, 2014, 113, pp. 414– Sustain. Energy, 2015, 6, (3), pp. 850–862
421 [31] Koad, R., Zobaa, A., El-Shahat, A.: ‘A novel MPPT algorithm based on
[13] Ahmed, R., Namaane, A., M'Sirdi, N.: ‘Improvement in perturb and observe particle swarm optimisation for photovoltaic systems’, IEEE Trans. Sustain.
method using state flow approach’, Energy Proc., 2013, 42, pp. 614–623 Energy, 2017, 8, (2), pp. 468–476. Available at http://dx.doi.org/10.1109/
[14] Kimball, J.W., Krein, P.T.: ‘Discrete-time ripple correlation control for TSTE.2016.2606421
maximum power point tracking’, IEEE Trans. Power Electron., 2008, 23, (5), [32] Daraban, S., Petreus, D., Morel, C.: ‘A novel {MPPT} (maximum power
pp. 2353–2362 point tracking) algorithm based on a modified genetic algorithm specialized
[15] Miyatake, M., Inada, T., Hiratsuka, I., et al.: ‘Control characteristics of a on tracking the global maximum power point in photovoltaic systems affected
Fibonacci-search-based maximum power point tracker when a photovoltaic by partial shading’, Energy, 2014, 74, pp. 374–388
array is partially shaded’. Proc. Fourth IPEMC, Xi'an, China, 2004, pp. 816– [33] Seyedmahmoudian, M., Mekhilef, S.M., Rahmani, R., et al.: ‘Maximum
821, 2 power point tracking of partial shaded photovoltaic array using an
[16] Nguyen, T.L., Low, K.: ‘A global maximum power point tracking scheme evolutionary algorithm: a particle swarm optimization technique’, J. Renew.
employing DIRECT search algorithm for photovoltaic systems’, IEEE Trans. Sustain. Energy, 2014, 6, (2), pp. 1–13, 023102 doi:10.1063/1.4868025
Ind. Electron., 2010, 57, (10), pp. 3456–3467 [34] Lian, K., Jhang, J., Tian, I.: ‘A maximum power point tracking method based
[17] Kuei-Hsiang, C., Ching-Ju, L.: ‘An intelligent maximum power point tracking on perturb-and-observe combined with particle swarm optimization’, IEEE J.
method based on extension theory for PV systems’, Expert Syst. Appl., 2010, Photovoltaics, 2014, 4, (2), pp. 626–633
37, pp. 1050–1055 [35] Sundareswaran, K., Kumar, V.V., Palani, S.: ‘Application of a combined
[18] Tafticht, T., Agbossou, K., Doumbia, M., et al.: ‘An improved maximum particle swarm optimization and perturb and observe method for {MPPT} in
power point tracking method for photovoltaic systems’, Renew. Energy, 2008, {PV} systems under partial shading conditions’, Renew. Energy, 2015, 75, pp.
33, pp. 1508–1516 308–317
[19] Heydari-doostabad, H., Keypour, R., Khalghani, M., et al.: ‘A new approach [36] Bataineh, K., Taamneh, Y.: ‘Adaptive neuro-fuzzy inference system-based
in MPPT for photovoltaic array based on extremum seeking control under improvement of perturb and observe maximum power point tracking method
uniform and non-uniform irradiances’, Sol. Energy, 2013, 94, pp. 28–36 for photovoltaic systems’, Int. J. Power Electron. Drive Syst. (IJPEDS), 2017,
[20] Noguchi, T., Togashi, S., Nakamoto, R.: ‘Short-current pulse-based 8, (3), pp. 1327–1334
maximum-power-point tracking method for multiple photovoltaic and [37] Farahat, M., Enany, M., Nasr, A.: ‘Assessment of maximum power point
converter module system’, IEEE Trans. Ind. Electron., 2002, 49, (1), pp. 217– tracking techniques for photovoltaic system applications’, J. Renew. Sustain
223 Energy, 2015, 7, p.042702
[21] Masoum, M.A.S., Dehbonei, H., Fuchs, E.F.: ‘Theoretical and experimental [38] Santos, L.J.L., Antunes, F., Chehab, A., et al.: ‘A maximum power point
analyses of photovoltaic systems with voltage and current-based maximum tracker for PV systems using a high performance boost converter’, Sol.
power-point tracking’, IEEE Trans. Energy Convers., 2002, 17, (4), pp. 514– Energy, 2006, 80, (7), pp. 772–778
522 [39] Liu, X., Lopes, L.A.C.: ‘An improved perturbation and observation maximum
[22] Kazmi, S., Goto, H., Ichinokura, O., et al.: ‘An improved and very efficient power point tracking algorithm for PV arrays’. Power Electronics Specialists
MPPT controller for PV systems subjected to rapidly varying atmospheric Conf. (PESC 04), Aachen, Germany, 2004
conditions and partial shading’. Proc. Australasian Universities Power [40] Kharb, R., Ansari, M.D., Shimi, S.: ‘Design and implementation of ANFIS
Engineering Conf. (AUPEC), Adelaide, SA, Australia, 2009, pp. 1–6 based MPPT scheme with open loop boost converter for solar PV module’,
[23] Miyatake, M., Veerachary, M., Toriumi, F., et al.: ‘Maximum power point Int. J. Adv. Res. Electr. Electron. Instrum. Eng., 2014, 3, pp. 2320–3765
tracking of multiple photovoltaic arrays: a PSO approach’, IEEE Trans.
Aerosp. Electron. Syst., 2011, 47, (1), pp. 367–380

IET Power Electron., 2019, Vol. 12 Iss. 4, pp. 703-711 711

© The Institution of Engineering and Technology 2018