Solar Energy 159 (2018) 134–141

Contents lists available at ScienceDirect

Solar Energy
journal homepage: www.elsevier.com/locate/solener

Backstepping based non-linear control for maximum power point tracking in T

photovoltaic system☆
⁎
Naghmasha, , Hammad Armghana,1, Iftikhar Ahmadc, Ammar Armghanb, Saud Khanc,
Muhammad Arsalanc
a
School of Electrical Engineering, The University of Faisalabad, Faisalabad, Pakistan
b
Department of Electrical Engineering, Aljouf University, Aljouf, Saudi Arabia
c
School of Electrical Engineering and Computer Science, National University of Sciences and Technology, Islamabad, Pakistan

A R T I C L E I N F O A B S T R A C T

Keywords: The increasing energy demands, depleting fossil fuels and increasing global warming due to carbon emission has
MPPT arisen the need for an alternate, overall efficient and environment-friendly energy system. Solar energy is
Backstepping considered to be one of the most promising alternative energy sources, but it has the problem of low efficiency
Power converters due to varying environmental conditions. To increase its efficiency, a maximum power point tracking (MPPT)
Photo voltaic
algorithm is required to harvest maximum power from the Photovoltaic (PV) array. In this paper, a non-linear
backstepping controller is proposed to extract the maximum power from the PV system. A non-inverting buck-
boost converter is used as an interface between the load and the PV array. Reference voltages for the controller
are generated by a regression plane. Asymptotic stability of the system is verified through Lyapunov stability
analysis. The performance of the proposed controller is tested under MATLAB/Simulink platform. The simula-
tion results validate that the proposed controller offers fast and accurate tracking. Comparison with perturb &
observe and fuzzy logic controller is provided to show the performance of the proposed controller under abrupt
variation of the environmental conditions.

1. Introduction
The unstable oil prices, increasing energy demand and the recent
concern about the global warming has encouraged us to move towards
renewable and sustainable energy sources. Solar energy is considered to
be one of the most prominent renewable energy sources for power
generation (Dincer, 2011). In 2015, Solar Photo Voltaic (PV) had made
a recorded history of adding 55 Gigawatts (GW) to the global installed
capacity, making it a total of about 227 GW (Ren21, 2016).
PV Panels have low installation cost and are environmentally
friendly, but they have the problem of low-efficiency (Başoğlu et al.,
2015). To increase their efficiency (output power of PV/maximum
power of PV), they must operate at the maximum power point. A typical
PV system is shown in Fig. 1, which consists of:
1. PV module which generates electric energy from solar energy.
2. DC-DC converter to transfer the maximum power according to the
load requirements.
3. MPPT controller that generates the maximum power from the PV
with the help of DC-DC converter.
4. Electric Load.
As tracking the MPP is the most important part of a PV system,
intensive research work is being done in this particular area to develop
new and more efficient MPPT controllers. The characteristics curve of
the PV module is shown in Fig. 2. In almost every MPPT technique, we
have to trace the Vmpp voltage or Impp current at which the PV module
will supply maximum power. The MPP depends upon weather condi-
tions i.e. temperature and irradiance. There are various methods to
trace the MPP which are divided into three broad categories: conven-
tional techniques, population-based algorithms, and artificial in-
telligence (AI) techniques.
Perturb and observe (P & O) (Xiao and Dunford, 2004; Jain and
Agarwal, 2004) and Incremental Conductance (INC) (Hussein et al.,
1995; Reisi et al., 2013) are the most commonly used MPPT methods. In
P & O, the difference in power (dP = P1−P2 ) is checked at different

☆
This document is a collaborative effort.
⁎
Corresponding author.
E-mail addresses: 14mseenmr@seecs.edu.pk (Naghmash), 14mseeharmghan@seecs.edu.pk (H. Armghan), iftikhar.rana@seecs.edu.pk (I. Ahmad),
ammar.armghan@gmail.com (A. Armghan), 14mseeskhan@seecs.edu.pk (S. Khan), marsalan.msee15seecs@seecs.edu.pk (M. Arsalan).
1
Principal corresponding author.

http://dx.doi.org/10.1016/j.solener.2017.10.062
Received 20 August 2016; Received in revised form 20 October 2017; Accepted 21 October 2017
Available online 03 November 2017
0038-092X/ © 2017 Elsevier Ltd. All rights reserved.
Naghmash et al.
Fig. 1. Block diagram of MPPT.
Fig. 2. PV characteristics curve.
Fig. 3. Power at different voltage levels.
voltage levels (V1 & V2 ) shown in Fig. 3. The voltage is perturbed in ei-
ther direction. If dP > 0 , the voltage is perturbed in this direction, else
the perturbation is done in the opposite direction. In INC, change in
power with respect to voltage (dP / dV ) is checked. MPP is achieved
when the slope of the PV characteristic curve is zero i.e. dp / dv = 0 .
Comparison of the instantaneous conductance (I / V ) to the incremental
conductance (dI / dV ) is done in this method (Brambilla et al., 1999). At
MPP, dI / dV = −I / V . If dI / dV > −I / V , we are at the left of MPP, so we
have to perturb the voltage to the right side. If dI / dV < −I / V , we are at
the right of the MPP, so we have to perturb the voltage in left direction.
The main problem with these conventional methods is that at MPP, the
perturbation doesn’t stop and there are oscillations around the MPP
which results in the loss of power.
Artificial intelligence based MPPT techniques include fuzzy logic
controller (FLC) and neural network (NN). FLC doesn’t need a
135
Solar Energy 159 (2018) 134–141
mathematical model that’s why it can handle non-linearities (El Khateb
et al., 2014). FLC based MPPT has good performance in varying irra-
diance levels, but it needs a good rule base table (Messai et al., 2011). In
Guenounou et al. (2014), the author has proposed an adaptive FLC,
which integrates into two different rules. The first one adjusted the duty
cycle of the power converter while the second one adjusts the gain of
the controller. It is shown that the adaptive FLC outperforms the con-
ventional one. In Liu et al. (2013), an artificial neural network (ANN)
based MPPT technique is proposed. The proposed controller has ad-
vantage of low computation requirement and fast tracking speed irre-
spective of the PV module used. In Kharb et al. (2014), author proposed
adaptive neuro-fuzzy inference system (ANFIS) based MPPT technique.
Temperature and Irradiance levels are taken as inputs to train the
ANFIS. It is shown that the ANFIS works well under varying tempera-
ture and irradiance levels. To be robust, the AI based MPPT techniques
require a massive database which needs long computation time and
large memory size.
Particle swamp optimization (PSO), Genetic Algorithm (GA) and
Ant Colony Optimization (ACO) are the commonly used population-
based MPPT techniques. In Ishaque et al. (2012) author has proposed a
PSO based MPPT algorithm. It is shown that the proposed algorithm has
slow oscillations around the MPP and can find the MPP under varying
environmental conditions. A modified GA-based MPPT technique is
proposed in Daraban et al. (2014). The proposed algorithm has been
integrated with P & O, which results in less iterations and small popu-
lation size. In Jiang et al. (2013), ACO based MPPT technique is pro-
posed which shows fast convergence in varying weather conditions. It is
shown that the proposed technique works better than PSO and P & O.
Population-based MPPT techniques are complex and require many
parameters such as population size, selection of chromosome, mutation
and crossover rate. These parameters need to be readjusted during
varying environmental conditions or else MPP cannot be tracked down.
In this paper, a nonlinear backstepping controller for MPPT is pro-
posed. The proposed controller uses regression plane to generate a re-
ference voltage (Vref ) and tracks the MPP using non-inverted buck-boost
converter. The non-inverted buck-boost converter is selected because of
its high capability for MPPT (Başoğlu and Çakır, 2016). The non-linear
controller is chosen due to the non-linear nature of the non-inverted
buck-boost converter for global stability (El Fadil and Giri, 2007).
Asymptotic stability is guaranteed using Lyapunov stability analysis
and MPP is achieved under varying temperature and irradiance levels.
The paper is structured as follows. Section 2 presents the modeling
of the PV array. Modeling of the non-inverted buck-boost converter is
shown in Section 3. In Section 4 the regression plane and analysis of the
proposed controller is done. Simulation results are explained in Section
5. Comparison of the proposed controller with perturb and observe
technique and fuzzy logic controller is showed in Section 6. Section 7
includes the conclusion and future work.
2. Modeling of PV module
A practical model of PV module is shown Fig. 4. This model is
known as single diode model (Villalva et al., 2009). It consists of:
1. Current source Ipv , which depends on irradiance and temperature.
Fig. 4. One diode model of PV module.
Naghmash et al. Solar Energy 159 (2018) 134–141

2. Diode, D.
3. Rp , parasitic resistance.
4. Rs , series resistance.

Using Kirchhoff current law in Fig. 4, we can write

I = Ipv−Id−IRp (1)
where I is the PV module output current, Ipv is current generated by PV
cell and is directly proportional to the irradiance, Id is the diode current
and IRp is the current through shunt resistance. From Josephs (1976),
Fig. 5. Circuit diagram of non-inverted buck-boost converter.
we can write Ipv, Id, IRp as

G ⎞ It is assumed that the converter is operating in continuous con-

Ipv = ⎛ ⎜ (Ipv,STD + K (T −TSTD ))
⎟

⎝ GSTD ⎠ (2)
where G is the current irradiance level, GSTD is the irradiance at stan-
dard conditions, Ipv,STD is the current generated by solar cell at GSTD and
TSTD, K is the Boltzmann constant, T is the p-n junction temperature and
TSTD is the standard p-n junction temperature.
q (V + Rs I ) ⎞
Id = I0 ⎜⎛exp −1⎟
⎝ γkT ⎠ (3)
where I0 is the diode reverse saturation current which can be re-
presented by the following equation (Rashid, 2009),
duction mode. There are two operating modes of the converter i.e.
mode 1, in which the switches are on and mode 2, in which the switches
are off.
In mode 1, both IGBT switches (S1 & S2 ) are ON, diode (D1) is reverse
biased and load is disconnected due to the closed path by switch (S2 ).
Inductor (L) is charged from PV through switch (S1) in this mode.
Using Kirchhoff’s voltage and current law, we can write
⎧i ci = ipv−iL
dV i
⎨ dtpv = Cpv − CiL
⎩ i i (8)

Ipv,STD + Ki (T −TSTD ) v = vpv

I0 = ⎧ L
vpv
exp ( qVoc + Kv (T − TSTD )
γVt )
−1
(4) ⎩
di
⎨ dtL = L (9)

where q is the electron charge, Voc is the PV open circuit voltage, Vt is −vo
⎧ic = R
the PV module thermal voltage, Ki is the short circuit coefficient, Kv is
⎨ dvo = −vo
the open circuit voltage coefficient and γ is the diode ideality factor. ⎩ dt RC (10)
V + Rs I In mode 2, Both switches (S1 & S2 ) are off and the load is connected
IRp =
Rp (5) to inductor (L) through diode (D2 ).
Using Kirchhoff’s laws, we can write
Putting values of Ipv, Id and IRp from Eqs. (2), (3) and (5) in Eq. (1),
we get ⎧i ci = ipv
dv ipv
G ⎞ q (V + Rs I ) ⎞ V + Rs I ⎨ dpv = C
I=⎛ ⎜ (Ipv,STD + Ki (T −TSTD ))−I0 ⎜⎛exp
⎟ −1⎟− ⎩ t i (11)
⎝ GSTD ⎠ ⎝ γkT ⎠ Rp
v = −vo
(6) ⎧ L
diL −vo
By solving Eq. (6) for complete PV module, we get: ⎨ dt = L (12)
⎩
o v
⎛ ⎡ ⎛ V + R I ⎞ ⎤ ⎞ V + Rs I ⎧ic = iL− R
I = Np ⎜Ipv−I0 ⎢exp ⎜ N kT s ⎟−1⎥ ⎟−
⎜ ⎢ ⎜ s γ ⎟ ⎥⎟ Rp ⎨ dvc = iL v
− o
⎝ q ⎠ ⎦⎠ (7) ⎩ dt C RC (13)
⎝ ⎣
V is the output voltage, Rs is the series resistance due to metal contact From inductor volt-second balance and capacitor charge balance,
joining the PV cells, Rp is the parasitic resistance due to the p-n junction we can write
leakage current, Ns is the number of cells connected in series and Np is dvpv ipv i
⎧ = −u CL
the number of cells connected in parallel. ⎪ dt Ci i
diL vpv v v
=u + u Lo − Lo
⎨ dt L
3. Modeling of non-inverted buck-boost converter ⎪ dvc iL v i
= − o −u CL (14)
⎩ dt C RC

DC-DC converter is used to step up or step down a DC voltage u

(Erickson and Maksimovic, 2007). In the proposed technique, a non- vo = vin
1−u (15)
inverted buck-boost converter is used. Non-inverted buck boost con-
verter is used to shift the PV array output voltage (Vpv ) to our desired By assuming ideal power transfer i.e. Pi = Po , we can write
Vmpp by changing the duty cycle with the help of the backstepping (1−D)2
controller. The basic circuit diagram of the non-inverted buck-boost is Rpv = Ro
D2 (16)
shown in Fig. 5. It consists of the following main components.
For control design, we will average the model over one switching
1. Input voltage from PV (Vpv ). period. if x1 is the average value of vpv, x2 is the average value of iL, x3 is
2. Insulated-gate bipolar transistor (IGBT) switches (S1 & S2 ). average value of vc and μ is the average value of u, i.e.
3. Inductor (L). x = < vpv >
⎧ 1
4. Diodes (D1 & D2 ). ⎪ x2 = < iL >
5. Capacitor (Ci & C). ⎨ x3 = < vc >
6. Load (R). ⎪ μ = < u> (17)
⎩

136
Naghmash et al. Solar Energy 159 (2018) 134–141

In Eq. (24), taking x2 as a virtual control input and define V1 as

Lyapunov candidate function to check the convergence of the error e1 to
equilibrium point.
1 2
V1 = e1
2 (25)
In order to assure the asymptotic stability, the Lyapunov function
must be positive definite and radially unbounded and its derivative
with respect to time must be negative definite (Khalil and Grizzle,
1996). Taking time derivative of Eq. (25), we get
V1̇ = e1 e1̇ (26)

Ipv x
V1̇ = e1 ⎛ −μ 2 −Vpvr
⎜ ̇ ⎞ ⎟

⎝ Ci Ci ⎠ (27)
Fig. 6. Maximum power point reference voltage for varying irradiance and temperature
levels.
For V1̇ to be negative, Let
Ipv x2
−μ ̇ = −k1 e1
−Vpvr
Then the Eq. (14) take the form Ci Ci (28)
Ipv x2 Ipv
x1̇ = −μ x2 = ⎛k1 e1 + ̇ ⎞ Ci
−Vpvr
Ci Ci (18) ⎜

Ci
⎟

⎝ ⎠μ (29)
x x + x3 using the values of x2 from Eqs. (29), (27) becomes
x2̇ = − 3 + μ ⎛ 1 ⎞
L ⎝ L ⎠ (19)
V1̇ = −k1 e12 (30)
x x x
x3̇ = 2 − 3 −μ 2 For V1̇ to be negative definite, k1 must be positive definite and Eq.
C RC C (20)
(29) must be satisfied. β is the stabilization function, acts as reference
which is the required model of non-inverted buck boost converter. current for x2 .

4. Controller design and stability analysis ̇ Ci ) 1

β = (Ci k1 e1 + Ipv−Vpvr
μ (31)
The primary control objective is to track the PV module output In order to track x2 to β , an error e2 is defined as:
voltage Vpv to reference voltage Vref generated from regression plane
under varying environmental conditions. e2 = x2−β (32)

or
4.1. Regression plane
x2 = e2 + β (33)
The regression plane for the PV panel is shown in Fig. 6. It gives Differentiating Eq. (33), e1̇ becomes
reference voltage for the backstepping controller to achieve MPP under
Ipv e2 + β ⎞ ̇
varying temperature and irradiance levels. e1̇ = −μ ⎛
⎜ −Vpvr
⎟

The matrix of Vmpp for different temperatures (0–75 °C) and irra- Ci ⎝ Ci ⎠ (34)
diance levels (200–1200 W/m2) is derived using the PV module char-
Ipv β
acteristic curve. Using linear interpolation, the regression plane is de- e1̇ = ⎜ ̇ −μ ⎛ e2 ⎞
−μ ⎛ ⎞−Vpvr⎟ ⎜ ⎟

rived which provides us the reference voltage i.e. Vpvr for the controller. Ci ⎝ Ci ⎠ ⎝ Ci ⎠ (35)
The backstepping controller then tracks the Vpv to Vpvr to achieve the
e
MPP. The derived regression plane can be expressed through Eq. (21). e1̇ = −k1 e1−μ ⎛ 2 ⎞ ⎜ ⎟

⎝ Ci ⎠ (36)
Vpvr = 574.824−0.112500∗irradiance−2.0468∗ Temperature (21)
Using Eq. (36), Eq. (26) becomes

4.2. Backstepping control e

V1̇ = e1 e1̇ = e1 ⎛−k1 e1−μ 2 ⎞
⎜ ⎟

⎝ Ci ⎠ (37)
In order to extract the maximum power from the PV module, a non- e1 e2
linear backstepping controller is designed to track the PV module V1̇ = −k1 e12−μ
Ci (38)
output Voltage Vpv to Vmpp by controlling the duty cycle of the converter.
For this purpose, first of all we define the error signal. Differentiating Eq. (32) with respect to time, we get
e1 = Vpv−Vpvr (22) e2̇ = x2̇ −β ̇ (39)
where Vpvr is the reference voltage generated by the regression plane. By Differentiating Eq. (31) with respect to time. Using quotient rule, β ̇
converging the e1 to zero, we can get our desired result. Taking deri- becomes
vative of Eq. (22), we get
1
β̇ = ((k1 e1̇ Ci + Ipv̇ −V¨pvr Ci ) μ−μ̇ (k1 e1 Ci + Ipv−Vpvr
̇ Ci ))
̇ −Vpvr
e1̇ = Vpv ̇ (23) μ2 (40)

Using Eq. (18), we get Putting Eq. (36), we get

Ipv x2 1 ⎛ ⎛ μ ⎞ μ̇
e1̇ = −μ ̇
−Vpvr β̇ = ⎜k1 −k1 e1−
⎜ e2 Ci + Ipv̇ −V¨pvr Ci ⎞⎟ μ− (β )
⎟
Ci Ci (24) μ2 ⎝ ⎝ Ci ⎠ μ (41)
⎠

137
Naghmash et al. Solar Energy 159 (2018) 134–141

1 μ̇ Table 1
β̇ = (−k12 e1 Ci−k1 μe2 + Ipv̇ −V¨pvr Ci )− (β )
μ μ (42) PV array parameters.

1 μ̇ Parameter Value
β̇ = (−k12 e1 Ci + Ipv̇ −V¨pvr Ci )−k1 e2− (β )
μ μ (43) Maximum power 1555 W
Cells per module 72
Putting Eqs. (19) and (43) in Eq. (39)
Voltage at open circuit 165.8 V
x3 x + x3 1 μ̇ Current at short circuit 17.56 A
e2̇ = − + μ⎛ 1 ⎞− (−k12 e1 Ci + Ipv̇ −V¨pvr Ci ) + k1 e2 + (β ) Voltage at maximum power 102.6 V
L ⎝ L ⎠ μ μ (44)
Current at maximum power 15.16 A
To insure the asymptotic stability of the system and the convergence
of the errors e1 and e2 to zero, a composite Lyapunov function Vc is
defined whose time derivative should be negative definite for all value simulations are performed in MATLAB/Simulink using
of x. “SimPowerSystems” toolbox under the abrupt variation of environ-
mental conditions. PV array has been connected to the load through
1 2 non-inverted buck boost converter controlled by the backstepping
Vc = V1 + e2
2 (45)
controller. The simulation results are divided into two subsections.
Taking its time derivative, we get Firstly, the simulations are done with varying irradiance levels, then
with varying temperature levels.
Vċ = V1̇ + e2 e2̇ (46)
The PV array used in this work consists of 16 PV modules i.e. 4
Using Eq. (38), we get modules are connected in series (Ns = 4 ) and 4 modules are connected
e1 e2 in parallel (Np = 4 ). There are 72 cells per module. The specification of
Vċ = −k1 e12−μ + e2 e2̇ the single PV module is shown in Table 1, whereas the specifications of
Ci (47)
the controller parameters and the non-inverting buck-boost converter is
e shown in Table 2.
Vċ = −k1 e12 + e2 ⎛e2̇ −μ 1 ⎞
⎜ ⎟

⎝ Ci ⎠ (48)
5.1. Test under varying irradiance
For Vċ to be negative, Let
e1 Under this test, temperature is kept constant at 25 °C and irradiance
e2̇ −μ = −k2 e2
Ci (49) is changed abruptly after every 0.1 s, i.e. with a perturbation frequency
of 10 Hz. The varying levels of irradiance is shown in Fig. 7. Initial
where k2 is constant and positive definite. Using values of ė2 from Eq.
irradiance is 650 W/m2 , then it is increased to 850 W/m2 at 0.1 s and
(44) in Eq. (49), gives
finally it is increased to 1000 W/m2 at 0.2 s.
x3 x + x3 1 ̇
μβ e Using PV array characteristic curves for each irradiance level, the
−k2 e2 = − + μ⎛ 1 ⎞− (−k12 e1 Ci + Ipv̇ −V¨pvr Ci ) + k1 e2 + −μ 1
L ⎝ L ⎠ μ μ Ci performance of the proposed controller is verified. Vref for the varying
(50) irradiance levels is generated by the regression plane and is tracked by
the backstepping controller. It can be seen in Fig. 8, that the proposed
Using Eqs. (49) and (50), Eq. (48) can be rewritten as
controller successfully tracks the reference voltage.
x x + x3 1 PV array output power along with PV array power curves is shown
Vċ = −k1 e12 + e2 ⎡− 3 + μ ⎛ 1 ⎞− (−Ci k12 e1 + Ipv̇ −V¨pvr Ci ) ⎤
in Fig. 9 which shows that the MPP is successfully achieved with little
⎢
⎣ L ⎝ L ⎠ μ ⎥
⎦
oscillations. It is also shown that during the abrupt variation of the
⎡ ̇
μβ e1 ⎤
+ e2 k1 e2 + −μ irradiance, the controller performed well, reaching the MPP in almost
⎢
⎣ μ Ci ⎥
⎦ (51) 0.02 s.
Fig. 10 shows the convergence of error signal to zero under the
Vċ = −k1 e12−k2 e22 (52)
abrupt variation of irraiance at 0.1 s and 0.2 s.
Solving Eqs. (51) and (52) for μ̇ ,
5.2. Test under varying temperature
μ 1
μ̇ = − ⎛⎜k2 e2− (−Ci k12 e1 + Ipv̇ −V¨pvr Ci ) + k1 e2⎞⎟
β⎝ μ ⎠ In this case, irradiance is kept constant at 1000 W/m2 while tem-
μ ⎛ x3 x1 + x3 e1 ⎞ perature is abruptly varied with the perturbation frequency of 10 Hz i.e.
− − + μ⎛⎜ ⎞−μ ⎟

β⎝ L ⎝ L ⎠ Ci ⎠ (53) after every 0.1 s. Initial temperature is set at 25 °C , then it is stepped up

to 45 °C at 0.1 s and 65 °C at 0.2 s. Fig. 11 shows the varying levels of the
1⎛ x x + x3 temperature.
μ̇ = −k2 e2 μ + μ 3 −μ2 ⎛ 1 ⎞−Ci k12 e1 + Ipv̇ ⎞
β⎝ L ⎝ L ⎠ ⎠ Using PV array characteristic curves, the performance of the pro-
1⎛ ¨ e posed controller is again verified. Regression plane provides the
+ −Vpvr Ci−k1 e2 μ + μ2 1 ⎞
⎜ ⎟

β⎝ Ci ⎠ (54)
Table 2
μ2 Simulation parameters.
1 x
μ̇ = ⎛⎜e2 (−k1−k2 ) μ−e1 ⎛Ci k12− ⎞ + μ 3 ⎞⎟
⎜ ⎟

β⎝ ⎝ Ci ⎠ L⎠ Parameter Value
1⎛ ̇ ¨ x + x
+ Ipv−Vpvr Ci−μ2 ⎛ 1 3 ⎞
⎞ Constant, k1 100
β⎝ ⎝ L ⎠⎠ (55) Constant, k2 9000

where 0 < μ < 1 and β ≠ 0

Input Capacitor, Ci 1 × 10−3 F
Inductor, L 20 × 10−3 H
Capacitor, C 48 × 10−3 F
5. Simulation results Load resistor, R 50 Ω
IGBT switching frequency, fs 5000 Hz
To verify the performance of the proposed controller, different

138
Naghmash et al. Solar Energy 159 (2018) 134–141

Fig. 7. Varying levels of irradiance. Fig. 11. Varying levels of temperature.

Fig. 8. Tracking of Vpv with respect to Vref under varying irradiance.

Fig. 12. Tracking of Vpv with respect to Vref under varying temperature.

Fig. 9. PV array output power under varying irradiance.

Fig. 13. PV array output power under varying temperature.

Fig. 10. Conversion of error signal to zero under varying irradiance.

optimum Vref for the backstepping controller, which tracked the re-
ference voltage, shown in Fig. 12. PV array output power along with Fig. 14. Conversion of error signal to zero under varying temperature.

reference power curves is shown in Fig. 13, which shows that the MPP
is successfully achieved in around 0.04 s with little oscillations. During

139
Naghmash et al. Solar Energy 159 (2018) 134–141

Fig. 15. Comparison of the proposed controller with P & O during varying irradiance. Fig. 17. Comparison of the proposed controller with P & O during varying temperature.

the abrupt variation of temperature at 0.1 s and 0.2 s, the controller

performed well, showing the robustness of the controller.
Fig. 14 shows the convergence of error signal to zero under the
variation of temperature at 0.1 s and 0.2 s.

6. Comparison with other MPPT controllers

To show the performance of the backstepping controller, it is

compared with P & O and FLC. Both controllers are simulated and
compared under varying temperature and irradiance levels.

6.1. Comparison with P & O Fig. 18. Zoomed view of the MPPT of both controllers during varying temperature.

The comparison of the proposed controller and the P & O is done in

two tests. In the first phases, the comparison is done with varying ir-
radiance while in second phase, the comparison is done with varying
with temperature.

6.1.1. Comparison under varying irradiance

In this comparison test, initially the irradiance is set at 700 W/m2
and is stepped up to 1000 W/m2 at 0.3 s. Fig. 15 shows the performance
comparison of P & O with the backstepping controller. It can be seen
that not only the proposed controller reach the MPP more rapidly than
the P & O but during the variation of irradiance at 0.3 s, the back-
Fig. 19. Zoomed view of the oscillations of both controller around the MPP.
stepping controller reach the MPP rapidly with minimal oscillations.
To see the comparative behavior after the variation of irradiance at
0.3 s, Fig. 16 provides the zoomed view of the MPPT of the both con-
troller.

6.1.2. Comparison under varying temperature

In the second comparison test, both controllers are tested done
under varying temperature levels. The irradiance is kept constant at
700 W/m2 and temperature is stepped down from 50 °C to 25 °C at 0.3 s,
shown in Fig. 17.
In Fig. 18, it can be seen that the proposed controller reach the MPP
in minimal time with less oscillations as compared to P & O.
A zoomed view of the oscillations around MPP of the both con-
trollers is provided in Fig. 19.
Fig. 20. Backstepping Vs Fuzzy Logic Controller under varying irradiance.

6.2. Comparison with Fuzzy Logic Controller (FLC)

The performances of the proposed controller and fuzzy logic con-

troller under varying irradiance are compared in Fig. 20. The tem-
perature is kept at 25 °C while irradiance is varied. Initially, the irra-
diance is kept 1000 W/m2 and then it is reduced to 600 W/m2 after 0.05 s.
After 0.1 s, the irradiance is reduced to 200 W/m2 . Clearly, the ripples
shown in the output of proposed controller are much smaller than those
produced as a result of applying FLC.
Similarly, their performance is also compared under varying
Fig. 16. Zoomed view of the MPPT of both controllers during varying irradiance.

140
Naghmash et al.
Fig. 21. Backstepping Vs Fuzzy Logic Controller under varying temperature.
temperature. In this scenario, the irradiance is kept at 1000 W/m2 while
the temperature of the PV array is varied. Initially the temperature is
kept at 25 °C and after the interval of 0.05 s, it is increased to 45 °C.
Similarly, after 0.1 s, it is increased to 65 °C. The output power of the PV
array is shown in Fig. 21. Again, the ripples in the output of the PV
array with FLC are greater than the backstepping control.
7. Conclusions and future work
In this paper, a non-linear backstepping based controller has been
designed to track the MPP of a PV system. The PV array is connected to
the load through a non-inverted buck-boost converter. To obtain max-
imum power from PV array, duty cycle of the non-inverted buck-boost
converter is controlled through which the PV array output voltage is
tracked to the voltage reference generated by the regression plane.
Global asymptotic stability of the system is verified through Lyapunov
stability analysis. It is shown that the controller performed well under
the abrupt variation of environmental conditions showing the robust-
ness of the controller. The comparison with the perturb and observe
technique and fuzzy logic controller is made which showed that pro-
posed controller performed better during the variation in the irradiance
and temperature levels.
Although the proposed controller successfully achieved the desired
objectives, however, the controller depends upon the regression plane
for its reference values. The regression plane needs to be updated due to
the aging or any type of malfunction in the PV array. To avoid such
cases, further improvements can be done by interfacing the controller
to a line search algorithm like golden section search to achieve much
better results during any type of malfunction in the PV array. Based on
the prior scenario, our future work includes.
• Interfacing of controller with search algorithm like golden search
method.
• Practical implementation of the proposed controller on a real
system.
• Designing of an efficient inverter to inject the PV array output power
into AC electrical system.
141
Solar Energy 159 (2018) 134–141
• Hybrid energy storage system to maximise the efficiency of the
overall system by storing the excess energy from the PV array in
batteries and supply the energy during peak energy demands.
References
Başoğlu, M.E., Çakır, B., 2016. Comparisons of mppt performances of isolated and non-
isolated dc–dc converters by using a new approach. Renew. Sust. Energy Rev. 60,
1100–1113.
Başoğlu, M.E., Kazdaloğlu, A., Erfidan, T., Bilgin, M.Z., Çakır, B., 2015. Performance
̇
analyzes of different photovoltaic module technologies under izmit, kocaeli climatic
conditions. Renew. Sust. Energy Rev. 52, 357–365.
Brambilla, A., Gambarara, M., Garutti, A., Ronchi, F., 1999. New approach to photo-
voltaic arrays maximum power point tracking. In: Power Electronics Specialists
Conference, 1999. PESC 99. 30th Annual IEEE, vol. 2. IEEE, pp. 632–637.
Daraban, S., Petreus, D., Morel, C., 2014. A novel mppt (maximum power point tracking)
algorithm based on a modified genetic algorithm specialized on tracking the global
maximum power point in photovoltaic systems affected by partial shading. Energy
74, 374–388.
Dincer, F., 2011. The analysis on photovoltaic electricity generation status, potential and
policies of the leading countries in solar energy. Renew. Sust. Energy Rev. 15 (1),
713–720.
El Fadil, H., Giri, F., 2007. Backstepping based control of pwm dc-dc boost power con-
verters. In: 2007 IEEE International Symposium on Industrial Electronics. IEEE, pp.
395–400.
El Khateb, A., Rahim, N.A., Selvaraj, J., Uddin, M.N., 2014. Fuzzy-logic-controller-based
sepic converter for maximum power point tracking. IEEE Trans. Ind. Appl. 50 (4),
2349–2358.
Erickson, R.W., Maksimovic, D., 2007. Fundamentals of Power Electronics. Springer
Science & Business Media.
Guenounou, O., Dahhou, B., Chabour, F., 2014. Adaptive fuzzy controller based mppt for
photovoltaic systems. Energy Convers. Manage. 78, 843–850.
Hussein, K., Muta, I., Hoshino, T., Osakada, M., 1995. Maximum photovoltaic power
tracking: an algorithm for rapidly changing atmospheric conditions. IEE Proc.-Gener.
Transm. Distrib. 142 (1), 59–64.
Ishaque, K., Salam, Z., Amjad, M., Mekhilef, S., 2012. An improved particle swarm op-
timization (pso)–based mppt for pv with reduced steady-state oscillation. IEEE Trans.
Power Electron. 27 (8), 3627–3638.
Jain, S., Agarwal, V., 2004. A new algorithm for rapid tracking of approximate maximum
power point in photovoltaic systems. IEEE Power Electron. Lett. 2 (1), 16–19.
Jiang, L.L., Maskell, D.L., Patra, J.C., 2013. A novel ant colony optimization-based
maximum power point tracking for photovoltaic systems under partially shaded
conditions. Energy Build. 58, 227–236.
Josephs, R., 1976. Solar Cell Array Design Handbook, NASA.
Khalil, H.K., Grizzle, J., 1996. Nonlinear Systems, vol. 3 Prentice Hall, New Jersey.
Kharb, R.K., Shimi, S., Chatterji, S., Ansari, M.F., 2014. Modeling of solar pv module and
maximum power point tracking using anfis. Renew. Sust. Energy Rev. 33, 602–612.
Liu, Y.-H., Liu, C.-L., Huang, J.-W., Chen, J.-H., 2013. Neural-network-based maximum
power point tracking methods for photovoltaic systems operating under fast changing
environments. Solar Energy 89, 42–53.
Messai, A., Mellit, A., Pavan, A.M., Guessoum, A., Mekki, H., 2011. Fpga-based im-
plementation of a fuzzy controller (mppt) for photovoltaic module. Energy Convers.
Manage. 52 (7), 2695–2704.
Rashid, M.H., 2009. Power Electronics: Circuits, Devices, and Applications. Pearson
Education India.
Reisi, A.R., Moradi, M.H., Jamasb, S., 2013. Classification and comparison of maximum
power point tracking techniques for photovoltaic system: a review. Renew. Sust.
Energy Rev. 19, 433–443.
Ren21, 2016. Renewables 2016 Global Status Report. < http://www.ren21.net/wp-
content/uploads/2016/06/GSR_2016_Full_Report_REN21.pdf > .
Villalva, M.G., Gazoli, J.R., Ruppert Filho, E., 2009. Comprehensive approach to mod-
eling and simulation of photovoltaic arrays. IEEE Trans. Power Electron. 24 (5),
1198–1208.
Xiao, W., Dunford, W.G., 2004. A modified adaptive hill climbing mppt method for
photovoltaic power systems. In: Power Electronics Specialists Conference, 2004.
PESC 04. 2004 IEEE 35th Annual, vol. 3. IEEE, pp. 1957–1963.