Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

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

Comparative and comprehensive review of maximum power point tracking

methods for PV cells
⁎
M.A. Danandeh, S.M. Mousavi G.
School of Railway Engineering, Iran University of Science and Technology, Tehran, Iran

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

Keywords: The energy problem is one of the most important and serious problems that humanity is faced with it and this is
Maximum power point tracking while the fossil fuels are running out, so finding new sources of energy is one of the challenges of modern man.
Photovoltaic Solar energy is an available, newable and almost eternal energy which can be converted directly to electrical
PV cell energy by photovoltaic (PV) cells. Although the use of sunlight costs nothing but PV cells are relatively expensive
so it's necessary to extract maximum power from these cells because of economic reasons. To achieve the
maximum power point, there are many techniques and also many review papers but just few papers have
compared these techniques from economical and technical point of view. This paper presents a review of MPPT
techniques using of comprehensive and relatively new classification with emphasizing on comparison of
methods.

1. Introduction N-type is located in that part of cell which is in front of sun and the P-
type is in the back of cell [2]. Also these cells have non-linear I-V and P-
The rapid growth of population and industry, technological progress V characteristics which both highly depend to irradiation, ambient
and improvement of life quality have led to an ever-increasing human temperature [3] and load impedance [4]. Figs. 1 and 2 show the non-
need for energy so that there is a direct relation between development linearity of the I-V and P-V characteristics of PV cell and effect of am-
and consuming of energy. Nowadays the big part of energy comes out bient conditions on them. As shown in figures increasement of tem-
by using of fossil fuels but issues like the political and economic crisis, perature reduces maximum power of PV cell while increment of irra-
environmental pollution, limitation of fossil resources and etc., have diance causes increment of it.
revealed the necessity of finding new sources of energy. Solar energy Today, the number of installed cells is rapidly increasing and these
can be used as one of the required sources for solving these problems. cells are used in three different types: stand alone, grid connected and
Solar energy is so clean and does not produce any greenhouse gases hybrid. For example more than 1.5 million photovoltaic systems have
which means environmental and health costs of generating energy will been installed in Germany [5] and generated power by them is about
decrease and also the global warming issue can be controlled so better 40,093 megawatts (MW) by the end of May 2016 [6]. Of course there
by using of renewable energy instead of fossil fuels. One of the other are still two important barriers in the path of PV cells, relatively ex-
benefits of solar energy is that it is worldwide and available such as pensive cost and low efficiency. Therefore it is completely necessary to
almost all countries can use it for generating energy without depending extract maximum power of cells because of economic reasons and
to other countries. Being eternal can be assumed as the biggest benefit growing demand for energy. So many methods have been presented for
of solar energy which guarantees the source can be used at least for tracking the maximum power point of PV cells and the most common of
thousands years so it can be counted as a livelong source. them have been reviewed in this paper.
Additional to the above mentioned, solar energy is free and com-
pletely safe with extremely much value of energy. The energy from the 2. Classification of maximum power point tracking( MPPT)
sun to the earth's surface is about one thousand times greater than the methods
energy that is released by fossil fuels during a day [1]. The use of PV
cells is one of the most common ways to exploit solar energy and di- There are many ways for classification MPPT methods. Four of the
rectly convert it to electrical energy. PV cells are generally composed of most common ways are listed in next:
two types of semiconductors which are called as N type and P type. The

⁎
Corresponding author.
E-mail addresses: [email protected] (M.A. Danandeh), [email protected] (S.M. Mousavi G.).

http://dx.doi.org/10.1016/j.rser.2017.10.009
Received 21 November 2016; Received in revised form 24 July 2017; Accepted 3 October 2017
1364-0321/ © 2017 Elsevier Ltd. All rights reserved.

Please cite this article as: Danandeh, M.A., Renewable and Sustainable Energy Reviews (2017), http://dx.doi.org/10.1016/j.rser.2017.10.009
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

Fig. 2. a : P-V characteristic for 0, 25, 50 and 75 °C. b : P-V characteristic for 0.4, 0.6, 0.8
Fig. 1. a : I-V characteristics for 0, 25, 50 and 75 °C. b : I-V characteristics for 0.4, 0.6, 0.8 and 1 kw/m2.
and 1 kw/m2.

2.4. Classification according to tracking style

2.1. Classification according to control mode
It can be a good way for classification MPPT methods. According to
In this way, MPPT techniques are divided into two classes, Online this way, methods that have same idea for tracking MPP go to the same
(direct) and Offline (indirect). In the online class, tracker does not re- class. The main advantage of this way is that methods which are in
quire to have a priori knowledge about the PV cell while offline same group have some same advantages, drawbacks and working style.
methods can not track MPP without it. The main advantage of online This way has been utilized for classification of methods in this paper
methods is being independent of priori knowledge which means online which is shown in Fig. 3.
methods can work in different conditions.
3. MPPT methods review
2.2. Classification according to number of variables
3.1. Curve fitting
MPPT methods can belong to one of one-variable or two-variable
groups. One-variable group consists of methods that need to measure The main idea of this method is providing a polynomial which can
just one parameter like current, voltage or temperature but two-vari- approximate the P-V characteristic of PV cell. This purpose can be
able group includes of methods which need to measure two parameters achieved by testing cell under different conditions, collecting experi-
like current and voltage. It is clear that few numbers of variables means mental data (P,V and etc.) and using curve fitting techniques. The most
few numbers of sensors. common approximation for P-V characteristic of cell is cubic order
polynomial as (1) [7].
2.3. Classification according to result of tracking P = a*V 3+ b*V 2 + c*V + d (1)

In this way, there are two groups of methods, true MPPT methods The coefficients of this relation are strongly dependent on tem-
and not true. "True" group includes of methods which tracks maximum perature of cell so (1) can be rewritten as (2).
power point exactly while "not true" methods use approximation for P (V,Tcell) = a (Tcell)*V 3+b(Tcell) *V 2+c(Tcell)*V + d (Tcell) (2)
tracking MPP which causes good speed for these methods but the result
of them is some point with high power and close to MPP but it is not According to Fig. 4 at the MPP, power variation with voltage is zero,
exactly maximum power point. so it is possible to calculate voltage of MPP by (3)–(5).

2
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

Fig. 3. Classification MPPT methods.

P&O Curve Newton-
fitting Raphson

MP&O Look-up
Secant
table

Power Stepest
EPP
Perturbation matching decsent

Hill based
Numerical
Bisection
climbing methods
RCC
Forced Central
OCC
oscillation point
Current
Three point sweep
CV RFM
weight
DC link cap.
MPPT MPPT
Load based MRFM
Temprature
based
dP/dV
Differential
Beta based
FSCC dP/dt
Cell based
FOCV INC
State space
Conduction PIAINC
Fuzzy logic based
MINC
Parastic
ANN
capacitance
VSINC
Intelligent
Firefly Sliding
methods
IVSINC mode
Hybrid
PSO
methods

Genetic

Fig. 4. Power variation with voltage.

dP
=0
dV (3)
Fig. 5. Cubic and fourth order approximation.
By using of (2) in (3), relation (4) can be concluded.
3*a(Tcell)*V2+ 2*b(Tcell)*V + c (Tcell) = 0 (4) P (V,Tcell) = a (Tcell)*V4+b(Tcell)*V 3+c(Tcell)*V 2+d (Tcell)*V + e (Tcell)
(6)
−b (Tcell)± (b (Tcell))2 −3a (Tcell) c (Tcell)
Vmpp =
3a (Tcell) (5) Where a(Tcell) , b (Tcell) ,…, d (Tcell) is calculated by (7).

Another approximation is fourth order which is more accurate than

cubic, Fig. 5 shows the fitting of the curve by cubic and fourth order
polynomial [8]. Fourth order polynomial has been expressed by (6).

3
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

a(Tcell) = a 4 *Tcell 4 + a3*Tcell3 + a2 *Tcell 2 + a1 *Tcell + a0 b(Tcell)

= b4 *Tcell 4 + b3 *Tcell3 + b2 *Tcell 2 + b1 *Tcell + b0 c(Tcell)
= c4 *Tcell 4 + c3*Tcell3 + c2 *Tcell 2+c *Tcell + c0 d(Tcell)
= d 4 *Tcell 4 + d3 *Tcell3 + d2 *Tcell 2 + d1 *Tcell + d 0 e(Tcell)
= e4 *Tcell 4 + e3 *Tcell3 + e2 *Tcell 2 + e1 *Tcell + e0 (7)

NOCT −20
Tcell = G + Tambient
800 (8)

Where Tcell presents temperature of cell in Celsius, G is irradiance value

on cell surface, NOCT is nominal operation condition temperature
(temperature of cell in 800 w/m2 and 20 C for ambient temperature)
[9] and a4, a3,…,e0 are dependent on different factors like temperature,
cell technology and etc. In this method temperature is measured then
coefficients and Vmpp are calculated by a microcontroller. All of these
steps are repeated periodically because ambient conditions may
change.
Main advantages: Simplicity. Fig. 6. a : equivalent circuit with one diode. b : equivalent circuit with two diodes.
Main drawback : offline (require priori knowledge), need for high
capacity memory because of high number of calculations, low speed,
dependence of coefficients on ambient conditions, not true mppt, ac- methods are relatively expensive.
curacy of approximation depends on number of experimental samples.
3.4.1. Open methods
3.4.1.1. Newton-Raphson. This is one of the most common iterative
3.2. Look up table based method
approaches which is addressed in ref. [16–31]. This method
approximates the function f(x) by using of tangent to function at a
This method is based on searching among the pre-stored informa-
point on f(x), as shown in Fig. 7, and do this by just one initial guess
tion. In the method voltage and/ or current is calculated (or measured)
[32].
for maximum power point under different conditions and saved in the
memory of the controller. During the operation, conditions and voltage In (x ) = f ′ (x n )·(x − x n ) + f (x n ) (9)
(and/or current) of cell are measured and compared with stored data to
f (x (n))
achieve MPP [10–12]. Also it is possible to use open circuit voltage x (n+1) = x (n) −
instead of Vmpp to make this method easier. f ′ (x (n)) (10)
Main advantage: Simplicity of operational logic. The point xn+1 is the root of the tangent ln(x) to the function f(x) at
Main drawbacks: need for large number of calculation or mea- the point xn [33] and n is the index of iterate. By replacing V instead of
surement to support wide range of environmental conditions, need for x:
high capacity memory, not comprehensive (can't support all condi-
f (v (n))
tions), not high accuracy, need for unique table for any different cell V (n+1) = V (n) −
and different conditions, offline (require priori knowledge), low speed, f ′ (v (n)) (11)
high volume for wide range of conditions, needing for more than two Where f (v) is dP/dV, so this formula can be rewriten as (12).
types of sensors.
dP
dV
v = v (n)
V (n+1) = V (n) −
3.3. Power matching method d2P
v = v (n)
dV2 (12)
There is a so simple method for running the loads direct connected
Where, V and P are output voltage and power of cell. Eq. (12) will be
to PV arrays. In this method PV cells with specific characteristic which
repeated until stop condition (13) is met.
matches well with specific load is used to supply required power [13].
As it is clear priori knowledge about cell and load has so important role f (x n ) ≤ ε (13)
in this method nevertheless this method just can approximate MPP.
Where ε is the acceptable tolerance.
Main advantages: simplicity
Main advantages: simplicity, need for one initial guess, fast
Main drawbacks: need for priori knowledge about cell and load, for
any load it's necessary to change cell and configuration.

3.4. Numerical based methods

The members of this group operate by solving the equals extracted

from equivalent circuit of PV cell via numerical algorithms. This group
is divided into two main sub-group : "Open methods" and "Bracketing
methods". Open methods use initial guess(es) as the root(s) of equal(s)
while bracketing methods are based on the intersecting a specified in-
terval without needing for initial guess. The common equivalent cir-
cuits of a solar cell are shown by a current source located in parallel
with one [14] or two diodes [15] which shown in Fig. 6. It should be
noted that the most common model illustrated by Fig. 6-a.
Main advantage: simplicity, good accuracy and efficiency.
Fig. 7. Newton-Raphson method.
Main drawback: speed depends on initial guess(es), most of these

4
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

3.4.1.3. steepest descent. This method like any other numerical method,
is based on repetition. Ref. [36] gives the relation (17) for variable x.
x (k +1) = x (k ) − αk ∇f (xk ) (17)

Where αk is step size and −∇f (xk ) is the direction of the dexcent which
will be calculated by (18) and (19), respectively.
αk = arg minα f (xk − α∇f (xk )) (18)

∂f (x ) ∂f (x ) ∂f (x ) T
−∇f (k ) = −⎡ …⎤
⎣ ∂x 1 ∂x2 ∂x3 ⎦ (19)

By assuming power of cell as f(x) and voltage as x, f(x) can be re-

writen as P(v) so relation (20) can be used for tracking MPPT.
dP
Fig. 8. Newton – Raphson method error. v = v (k )
V (k +1) = V (k ) dV
Kε (20)
convergence and good accuracy. Where V(k) and P(k) are voltage and power of cell in instant k. K ε is
Main drawbacks: Speed dependence on initial guess, need for which called as step size corrector and defined by (21).
causes for expensive computation part and increasement of error
(Fig. 8). 1
Kε = −
αk (21)
3.4.1.2. Secant method. Newton-Raphson method is good and powerful
method for MPPT but it needs to have a second order derivative which 3.4.2. Bracketing methods
means it will be expensive and sometimes difficult to calculate. Secant 3.4.2.1. Bisection method. This is one of the strongest, simplest and
method has a similar concept to Newton-Raphson technique but this most primitive numerical methods which is called as Binary Chopping,
one does not require to second-order derivative, Fig. 9 shows the Bolzano's Method or Interval Halving alternatively [32]. Ref. [33] and
concept of this method. MPP can be approximated by using of two [37] review the algorithm of this method. By assuming that interval [a,
initial guesses (V0, V1) and relation (16) [34,35]. b] includes a root of f(x) which is called as x*, the midpoint of the
interval can be calculated by (22).
f (x (n)) − f (x (n−1))
hn (x ) = . (x − x (n)) + f (x (n))
x (n) − x (n−1) (14) a+b
ci =
2 (22)
x (n) − x (n−1) In the next step f(a), f(b) and f(c) will be calculated and :
x (n+1) = x (n) − f (x (n)).
f (x (n)) − f (x (n−1)) (15)
a) If f (a) . f (c) is less than zero, a root is in [a, c]. Set b = c and repeat
Where hn(x) is secant line, f(x) is desired function, xn+1 is the root of
previous steps.
hn(x) and xn is intersection point of hn(x) and f(x). By replacing V
b) If f (b) . f (c) is less than zero, a root is in [c, b]. Set a = c and repeat
instead of x and dP/dV as f(x), relation (16) can be extracted.
previous steps.
dP V (n) − V (n−1) c) If f (c) = 0, then the root is c.
V (n+1) = V (n) − ⎧ v = v (n) ⎫ .
⎨
⎩ dV ⎬
⎭ ( dP
dV ) (
v = v (n) −
dP
dV
v = v (n−1) ) The algorithm will stop when | f(c) | ≤ ε. Where ε is acceptable
(16) error (degree of accuracy). To apply this method into MPPT, f(x) can be
glanced as the change in power (ΔP), where the variable x is the voltage
In (16), V and P are output voltage power of cell.
of solar module so f(x) can be written as dP/dV.
Main advantages: fast convergence, good accuracy, high efficiency.
Main advantages: no require for derivatives of voltage, complexity
Main drawbacks: no guarantee of stability and convergence for in-
is less than Newton-Raphson method, easy generation of control signal,
itial guesses far from MPP.
guaranteed convergence.
Main drawbacks: quite slow.

3.4.2.2. Central point. In this method defined range for control

parameter is divided into four intervals without any common point
which it causes three center points as shown in Fig. 10-a. Powers of
these three points are compared and two intervals that more likely to
include MPP are picked out of four existent intervals. In the next step
these two intervals are divided into another four intervals and these
steps will be repeated until the MPP is achieved. Fig. 10-b shows these
steps [38].
Main advantages: so suitable for resonant converter, variable per-
turbation during tracking, fast speed, reducing the effects of several
peaks.
Main drawbacks: offline, expensive, complexity.

3.4.2.3. Regula falsi method (RFM). This method can be seen as a

combination of bisection search and secant methods. Some Refs. like
[39–41] explain this method and ref. [33] summarizes it in four steps.
Fig. 9. Secant method.
Steps of this linearly convergent root-finding algorithm have been listed

5
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

convergence, more stable and accurate than Newton-Raphson and bi-

section search.
Main drawbacks: Slower than Newton – Raphson, offline, medium
speed, expensive.
3.4.2.3.1. Modified regula falsi method. One of the main drawbacks
of regula falsi method (RFM) is its speed which is less than that of
Newton- Raphson method and if f(x) is a concave or convex, this
problem will intensify. For solving this problem ref. [33] presents
Modified regula falsi method (MRFM). This method has four steps, like
RFM.

a) The interval [xl, xu] includes the root of f(x), if xl ≠ xu and f(xl).f
(xu) < 0
b) If f(xl)·f(xu) < 0 and f(xl) > 0, then (24) is used to approximate the
root.
0.5.xl . f (x u ) − x u f (xl )
ci =
0.5.f (x u ) − f (xl ) (24)
If f(xl)·f(xu) < 0 and f(xl) < 0, then (25) will be used.
xl . f (x u ) − 0.5.x u. f (xl )
ci =
f (x u )−0.5.f (xl ) (25)

c) If | f (ci) |≤ ε (where ε is the tolerance), then it is considered that ci

is the root. if f(ci )·f(xu ) < 0, then set xl = ci and if f(ci )·f(xl ) < 0,
then let xu = ci.
d) Repeat steps (b) and (c) until the root is reached. It should be no-
ticed that function f(x) must be a continuous function.

3.4.2.4. Linearization. PV cell has non-linear I-V characteristic which

makes difficult tracking of MPP. Ref. [42] presents a method which its
main idea is based on the intersection of two equals (Eqs. (28) and (30))
and the introduction of their intersection point as MPP. The I-V
characteristic of PV cell can be expressed by:
q
I = Is − Io ⎡e AkT (V + IRs)−1⎤
⎣ ⎦ (26)
Where : Is is generated current, Io is the reverse saturation current, I is
output current, R is the internal resistance of the cell, K is the
Boltzmann's constant, T is absolute temperature, q is charge of one
electron, A is an ideality factor for a p-n junction and V is output
voltage of cell. By (27) the P-I characteristic of cell is defined as
function f(x).
f (P , I ) = 0 (27)

q Is + Io − I
f (P , I ) = P − VI = P − I ln ⎛ ⎞ − RS I 2 = 0
AKT ⎝ Io ⎠ (28)
Fig. 10. a : Central point iterations. b : Central point method flowchart.
At MPP the variation of P with I equals zero (like Fig. 4) which is
in next. represented by (29) (Fig. 11).
dP
g (P , I ) = =0
a) An interval like [xl, xu ] should be defined in order to include a root dI (29)
of function, so initial points xl and xu should be found such that
A*k*T
satisfy two conditions: xl ≠ xu and f(xl).f(xu) < 0. ⎛ q
I2 ⎞
g (P , I ) = P − a ln ⎜ +Rs I 2 = 0
b) The root will be approximated by "ci" as (23). ⎜ (P − Rs I 2) Io ⎟⎟
⎝ ⎠ (30)
xl f (x u ) − x u f (xl )
ci = Eq. (30) can be approximated by a linear line. Fig. 12 shows the
f (x u ) − f (xl ) (23)
approximations in this method.
Main advantages: simplicity, inexpensive, instantaneously tracking,
c) If | f (ci) |≤ε, then ci is the root, if f(ci)·f(xu ) is less than zero, then high speed and accuracy.
set xl = ci and if f(ci )·f(xl ) < 0, then let xu = ci. Main drawbacks : offline (require priori knowledge about PV cell
d) Steps (b) and (c) will be repeated until | f (ci) |≤ε. where ε is the parameters)
acceptable tolerance. For tracking MPP, dP/dV should be replaced
with f(x). 3.5. Cell based methods

Main advantages: faster than bisection method, guaranteed These methods operate by two important properties of PV cell, open

6
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

Fig. 11. a : f (a) and f(c) have opposite sign. b : f (b) and f(c) have opposite sign.

circuit voltage (Voc) and short circuit current (Isc). The main idea of
these methods is that in any condition there is just one MPP for each
array or cell means Vmpp and Impp are unique, so MPP can be tracked
by tracking Vmpp or Impp. These methods try to present a relation
between Vmpp and Voc (or between Impp and Isc). Two members of
cell based methods family are FSCC and FOCV.
Main advantages: Simplicity and fast speed.
Main disadvantages: Not true MPPT, requirement for knowledge
about physical parameters of cell.

3.5.1. Fractional short circuit current (FSCC) method

In refs. [43,44] relation between Impp and Isc has been presented as
(31).

Immp = Ki Isc (31) Fig. 12. a : P-I characteristic of cell (Eq. (24)). b : maximum output power curve and its
linear approximation. c : MPP is at intersecting point.
Where Ki is called as current factor and is always less than one, typically
between 0.85 and 0.95 [45]. Ki mainly depends on cell specifications, switch to make short circuit periodically and the result is increment in
temperature and irradiance so PV scanning is done for calculating Ki number of components and cost [7].
[46] and Isc periodically. Of course in the most of time Ki can be Main advantages : Good speed, simplicity, ability to work in both of
considered as a constant parameter. After calculating Impp by (31), analog and digital mode, need for current sensor only, being cheap.
current is sampled and compared with Impp and error is applied to Main drawbacks: Not true MPPT, offline (require to know physical
converter to change the duty cycle. Measuring Isc during operation has parameters of cell for calculating Ki), periodically loss of power during
a negative point and it is that during the short circuit, voltage is zero so measurements of Isc, additional switch for making short circuit.
no power will be supplied in these periods of time. Another problem
which is created by measurement of Isc, is the need for additional

7
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

Fig. 13. MFSCC flowchart.

3.5.1.1. Modified fractional short circuit current (MFSCC) unequal to zero, a flag will be set by the Ipv loop
method. Although FSCC is a simple and quick technique but there are In Ipv loop, the output current of cell (Ipv) is measured, then its
some problems with it. If Isc considered as a constant parameter by value is divided by Isc and answer is saved as K2. Now K2 is divided by
changes in ambient conditions, this method will not be a valid tracker K1 which is constant of normal loop. if answer is less than the limit,
and if Isc measured periodically (for example every two seconds), it will then IMPP will not be changed. If the difference between the two con-
cause increasement of power loss because during measurement of Isc, stants is greater than the limit, then the algorithm sets the flag as 1, so it
voltage is zero. For solving this problem, MFSCC has been presented in is necessary new value of Isc to be measured and saved. This new value
[45]. The change in temperature is slow so the effect of its change is less will be used in loops instead of previous value, for adjusting duty cycle.
than irradiance change effect. With regard to this issue, the effect of The flag will be cleared by the normal loop and Ipv loop will again
temperature change has been neglected in this method. Fig. 13 follow its routine. It should be noticed that value of limit defines the
illustrates the flowchart of the MFSCC. As seen this method has two system sensitivity so if this value is small, sensitivity of system and
loops: Normal loop and Ipv loop. Normal loop is as same as FSCC numbers of Isc measurement will increase which mean more power loss
method. In the first stage cell is separated from the system and is made and expensive hardware while the large value of limit means low sen-
short circuit to Isc be measured. The measured value is stored in the sitivity and few numbers of measurements but its drawback is that
memory and then Impp is calculated in same way of FSCC by (31). Ipv system may work far from MPP, so a tradeoff for detecting value of limit
is compared with Impp and error goes to controller for adjusting duty is necessary. A good way for detecting this value is trial and error.
cycle of converter, when the error is zero, operating point is MPP. All of
these steps will be repeated periodically. As soon as the error becomes

8
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

input. This method almost is applicable for all types of load. Loads can
be divided to four groups: voltage-source type, current-source type,
resistive type, or a combination of these. If load is voltage-source type
the current of it, if load is current-source type voltage of it and if load is
other type, voltage or current of it should be maximized.
Main advantages: one variable, simplicity (no multiplier), few
number of sensors, high accuracy and speed.
Main drawback: low efficiency

3.8. Differential based methods

Fig. 14. Flowchart of open circuit method.
This group consists of high accurate methods for tracking MPPT of
PV cell. The main idea of these methods is that power variation with
3.5.2. Fractional open circuit voltage (FOCV) method
parameters like voltage, current and time is zero at MPP.
This is so similar to fractional short circuit current method and its
Main advantages: online, real MPPT, high accuracy and efficiency.
flowchart is shown in Fig. 14. Refs. [47–50] present a linear relation
Main drawbacks: being expensive, low speed, difficult implement.
between Vmpp and Voc as (32).
Vmmp = Kv Voc (32) 3.8.1. Power variation with time method
Where Kv is voltage factor typically between 0.73 and 0.8 (for poly- According to refs. [62,63] power variation with time is zero at MPP
crystalline) [51–53] and in some references it is between 0.78 and 0.92 and relation (33) expresses it.
but in general, Kv depends on cell specifications, temperature and ir- dP
radiance. Kv is calculated by PV scanning and Voc is measured by dis- =0
dt (33)
connecting converter, and both should be done periodically. It is pos-
sible to assume Kv as a constant parameter for all conditions to make dP dVI dV dI
= = 0→I +V =0
this method simpler. The value of this constant is achieved experi- dt dt dt dt (34)
mentally by measuring VOC and VMPP under different conditions [54]. It
Where I, V and P are output current, voltage and power of cell, re-
should be noted that assuming Kv as a constant will reduce the accuracy
spectively, and t is time. MPP can be tracked by using of (34). This
of method. Disconnecting converter for measurement of VOC needs for
method has low speed and require expensive controller because of large
additional components and also causes a big problem and that is power
number of calculation, as seen in Eq. (34) at least eight calculations is
loss. To solve this problem there are two common ways, using pilot cells
necessary : I, V, dI/dt, dV/dt, I* dV/dt, V* dI/dt and (I* dV/dt) + (V*
[55] or modified fractional open circuit voltage (MFOCV) method.
dI/dt).
MFOCV is so similar to MFSCC.
Main advantages: online, real MPPT, independent of temperature
Main advantages: Good speed, simplicity, ability to work in both of
and irradiance.
analog and digital mode, need for voltage sensor only, being cheap,
Main drawbacks: difficult implement, large number of calculation,
suitable for small scale cheap applications.
low speed, long time, expensive, need for measurement of two para-
Main drawbacks: not true MPPT, offline (require to know physical
meters: voltage and current (two sensors).
parameters of cell for calculating Kv), periodically power loss during
measurement of Voc, need for additional cells for measurement of Voc,
It is difficult to choose an optimal value for the constant K. 3.8.2. Power variation with voltage or current
As shown in Fig. 4, power variation with voltage is zero at MPP and
3.6. Constant voltage or current (CV or CI) method it's true about power variation with current [64,65]. First of all dP/dV is
measured or calculated. Although calculation of power variation with
This can be the simplest way for tracking MPP. It's enough to cal- voltage has different methods like sampling and calculating by digital
equipment [66] or linearization [67] but the fundamental is same. If
culate or measure Vmpp (or Impp) beforehand and set its value as re-
ference. Controller will measure voltage (or current) and compare its dP/dV (or ∆P / ∆V ) is more than zero it means that P < Pmax and
V < Vmpp in other words current point is on the left side of MPP so the
value with reference and apply generated error to DC/DC converter
which causes changes in duty cycle of it. It's clear that because of power should be increased by increasing voltage. If dP/dV (or ∆P / ∆V )
is less than zero it means P < Pmax and V > Vmpp (current point is on
controlling one parameter, voltage or current, sensors are few in
number. This method is for battery less systems and also known as the right side of MPP) so the power should be increased by reducing
voltage. Fig. 15 illustrates it. By replacing i instead of V, power varia-
voltage (or current) feedback technique [56–59].
Main advantages: simplicity, few numbers of sensors, being eco- tion with current will be achieved.
nomical and more efficency rather than some common methods like
P & O and IC methods in low insulation conditions [60].
Main drawback: offline (require a priori knowledge), useable just
for one specific condition (condition which Vmpp or Impp has been
calculated for it) in other words it is not valid if condition changes, not
real MPPT, not applicable to systems with battery.

3.7. Load based method

The out coming power of PV cell will be maximum if the converter's

output power is maximum. Ref. [61] presents a MPPT method by using
of this simple concept. The main idea of this method is tracking max-
imum output power of converter, so unlike the most of tracking
Fig. 15. Power variation with voltage.
methods, this method uses output parameters of converter instead of its

9
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

3.9. Perturbation based methods inaccuracy in sudden changes of environment conditions and not being
a real time method for slow perturbation process. For solving or redu-
This group includes some of the most prevalent MPPT methods. cing these problems, modified P & O (MP & O) methods have been
These methods are based on applying perturbation and comparing the presented [90]. There are many modified P & O methods with different
result point with previous point. Although in these methods operating speed and accuracy. Ref. [91] presents a method which according to EN
point oscillates around MPP but they are the most common methods in 50530 dynamic efficiency test has notable improvement. The change in
commercial systems because of simplicity, accuracy and good speed. temperature is slower than change of irradiance so it is possible to as-
Main advantages: simplicity, accuracy and good speed. sume that change of conditions means change of irradiance. In this
Main drawback: not suitable in fast changing conditions, oscillating method, voltage and current of operating point are measured and the
around MPP, not global MPP tracker. power is calculated then a perturbation with defined size is applied to
voltage and the new power is calculated and compared with power of
previous point and at the same time, normalized power (Pn = ∆P/P) is
3.9.1. Perturbation and observation (P & O) method
calculated and compared with the threshold value (∆Tr ). If Pn is less
This is the most common method in commercial systems and also
than ∆Tr , it means that change of the power and also change of irra-
one of the most common methods in papers [68–82]. In first stage,
diance were in small intervals. In the such situation, the size of per-
voltage and current of PV cell are measured and power is calculated
turbation is kept constant. If Pn is greater than ∆Tr , it means reducing
(Pa) then a perturbation with defined size is applied to voltage (∆V) or
the intensity of radiation. In this situation, perturbation size should be
current (∆i) and new power is measured (Pb). Now Pa and Pb are
increased. Also in this method, two barriers are defined on two sides of
compared. If Pb is greater than Pa (∆P > 0 ), it means perturbation was
MPP to prevent divergence. From FOCV method, known that Vmpp is
in true direction so another perturbation will apply in this direction and
almost 0.8 of open circuit voltage so it can be fine to set lower limit of
(Pc) will be measured and compared with (Pb). If Pb is less than Pa
voltage as 0.6 of Voc.
(∆P < 0 ), it means perturbation was in wrong direction so another
Other problem of P & O is oscillation of operating point around the
perturbation will apply in reverse and (Pd) will be measured and
MPP. Offer of this method for solving this problem is that when oper-
compared. These steps will be repeated until MPP is achieved, so it is
ating point gets close to MPP enough, perturbation size is made so
clear that this is a trial and error method [83–89]. Fig. 16 shows
small. For example initial size of perturbation can be considered as 2%
flowchart of this method.
of Voc and after that operating point closed to MPP, the size is changed
Main advantages: online (not require to priori knowledge), simpli-
to 0.5% of Voc. It should be noticed that size of perturbation cannot be
city of logic, simplicity of implement, medium speed, medium or good
zero because temperature and irradiance have a little and continuously
accuracy (dependent on perturbation size)
fluctuate. Ref. [92] presents a multivariable P & O algorithm which
Main drawbacks: not suitable for fast changing conditions, defining
track MPP by perturbing more than one control variable and ref. [93]
of suitable size for perturbation is difficult and important, oscillating
reviews fixed size and variable size P & O methods for grid connected
around MPP, not GMMPT.
PV array.

3.9.2. Modified perturbation and observation method

P & O is one of the most common methods and also it's easy to im- 3.9.3. Estimated-perturb-perturb (E.P.P.) method
plement but it has some notable problems like oscillating around MPP, E.P.P. is another method based on perturbation which presented to

Fig. 16. P & O method flowchart.

10
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

Fig. 17. E.P.P. method.

improve the tracking speed of the MP & O method. Fig. 17 shows the input resistance of these types of converters depend on their switching
flowchart of this method. As seen it is the evolved form of MP & O and frequency which means the output resistance of array (input resistance
the difference is that in EPP method there is an additional stage which of the converter) can be made equal to the load resistance by adjusting
is called as estimate mode. The added stage causes improvement of switching frequency. Making output resistance of PV array equal to
performance and increasement of tracking speed. Tests prove that this input resistance of load is done by applying a perturbation into the
method is about 1.5 time faster than MP & O [90,94]. The advantages switching frequency. Fig. 18 shows equal circuit of SEPIC and Cuk
and drawbacks of this method is so similar to MP & O method but EPP is converters.
faster.
vi 2
Pi = Po =
3.9.4. Hill climbing ri (35)
Hill climbing is one other of the most common methods. This
method is so similar to P & O and the difference is in the parameter
which perturbation is applied to it. In this method, power is measured
then a perturbation is applied to duty cycle of converter (∆D ) and
power is measured again. If ∆P is greater than zero means perturbation
was in true direction and other perturbation is applied again in the
same direction. if ∆P is less than zero means direction was not right so a
perturbation should be applied in reverse direction. All of these steps
are done successively to get MPP [70,95–101].
Main advantages: online (not require to priori knowledge), simpli-
city of logic, simplicity of implement, medium speed, medium or good
accuracy (dependent on perturbation size).
Main drawbacks: not suitable in fast changing ambient conditions,
suitable size for perturbation is important, oscillating around MPP, not
GMMPT.

3.9.5. Forced oscillation method

Moritz von Jacobi published theory of maximum power transfer,
about 1840 which is called "Jacobi's law" today [102]. According to this
law to extract maximum power from a source, the resistance of the load
must be equal to the viewed resistance of the source at its terminals.
Ref. [103] explains a MPPT method based on Jacobi's law. This method
requires that the solar array output resistance is matched with the load
resistance by a device so a PWM SEPIC or Cuk converter (DC/DC) is
Fig. 18. a : SEPIC converter. b : CUK converter.
used to connect the PV array to the load. Reason of this selection is that

11
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

Where Pi, vi and ri are input power, input voltage and input resistance
of converter. Po is output power of PV cell and also output voltage and
resistance of cell are vg and rg.
v v2
∂Pi = 2 i ∂vi − i2 ∂ri
ri ri (36)
At MPP : ∂Pi = 0 and according to Jacobi’s law ri = rg so :
δvi ∂v Vi
≈ i =
δri ∂ri 2rg (37)
Where Vi is the average of input voltage. At the end for discontinuous
inductor current mode :
ˆ sin(2π (fm) t )
fs = fs + fs (38)

1−k 2
ε1(k ) = β
(1+k 2)2 (39)
Where fs is switching frequency after variation applying, ε1(k ) is error,
ˆ is the maximum frequency
fs is the nominal frequency of switching, fs Fig. 20. Nine different possible states of the three perturbation points.
deviation, fm is the modulating frequency and is much lower than fs , D
is duty cycle, DMPP is require duty cycle at MPP, k = D/Dmpp and
ˆ
fs 3.9.6. Three point weight comparison method
β= 2fs
. This is another form of P & O method which compares three points
instead of two. A perturbation periodically applies to voltage of solar
⎧ if D < DMPP ε1(k ) > 0
if D = DMPP ε1(k ) = 0 array's terminal and then output power of array are compared on three
⎨ points. These three points are: the current point (A), a point perturbed
⎩ if D > DMPP ε1(k ) < 0 (40)
from (A) which can be called as (B) and a point with opposite direction
and for discontinuous capacitor voltage mode : perturbed from point (A) called (C). If power of point (B) is greater than
or equals to power of point(A), a positive sign will mark this state and if
k ′2 (k ′2−1) power of (B) is less than power of (A), a negative sign will be generated.
ε 2(k′) = β
(1+k ′2)2 (41) If the power of point (C) is less than power of point (A), a positive sign
Where k′ = (1-D)/(1-DMP) will be generated while a negative sign shows that power of (C) is
bigger than power of (A). After comparison, if two positive signs exist,
⎧ if D < DMPP ε 2(k′) > 0 the duty cycle of the converter should be increased (voltage should be
if D = DMPP ε 2(k′) = 0 increased as big as perturbation size) to move from (A) to (B), while two
⎨
⎩ if D > DMPP ε 2(k′) < 0 (42) negative signs mean that duty cycle of the converter should be de-
creased (moving from (A) to (C) ). If one positive and one negative sign
The method utilizes (40) and (42) for tracking MPP. If ε1(k ) > 0 or exist, the MPP is achieved or the change of solar radiation was too fast
ε 2(k′) > 0 , it means that duty cycle (D) should be increased. If ε1(k ) < 0 or [104]. Fig. 20 shows the nine possible cases. Main advantages and
ε 2(k′) < 0 , duty cycle should be decreased and if ε1(k ) or ε 2(k′) is equal drawbacks of this method are similar to P & O method but this one is
to zero, duty cycle should be kept constant. The implemented circuit of faster.
this method is seen in Fig. 19.
Main advantages: no oscillation around MPP, high accuracy and
good efficiency. 3.10. Conduction based methods
Main drawbacks: being expensive and complex, not being fast (it
has medium speed). These methods are some of the fastest and most accuracy methods

Fig. 19. Circuit of forced oscillating method.

12
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

for tracking MPP. The fundamental concept of this group is same with so it might not track the exactly MPP and will oscillate around MPP in
differential methods and the group is divided into three main methods, large confine, on the other hand small size causes increasement of ac-
Incremental conduction (Inc. Cond.), Parasitic capacitance and Sliding curacy but will lead speed to low, so tradeoff is necessary [7,115]. In-
mode. cremental conduction is so powerful and accurate method but it has
Main advantages: good speed, good accuracy, low oscillation. some disadvantages which the most important of them are increment
Main drawbacks: complexity, sensing two parameters (V,I) size problem and being unsuccessful in tracking global MPP, for solving
these problems modified INC method and some other methods have
3.10.1. Incremental conduction (INC) method been presented.
INC is one of the most common methods in commercial systems Main advantages: online, being fast, very good accuracy, adapt-
because of high speed and accuracy and also adaptability in fast ability to fast changing conditions, oscillation around MPP is less than
changing conditions. This method has been explained or implemented P & O method.
in many references like [105–114]. According to Fig. 15, dP/dV is zero Main drawbacks: dependence of speed and accuracy on increment
at MPP, more than zero on the left of MPP and less than zero on the size, the likelihood of having considerable oscillation, need for ex-
right of MPP. pensive controller, Require to two kind sensors (V and I), long response
time in changing conditions.
dP
⎧ dV < 0 on the right of MPP
⎪ dP 3.10.1.1. Modified Incremental conduction (MINC) method. This method
=0 at MPP
⎨ dV has been presented for solving the GMPPT problem in INC method. Ref.
⎪ dP > 0 on the left of MPP [116] explains it for using under partial shading conditions. The main
⎩ dV (43)
idea of this method is to find local peaks and choosing the highest of
dP d (VI ) dI ∆I them as GMPP. Step size of change in converter's duty cycle is defined
= =I+V ≅I+V
dV dV dV ∆V (44) as the control parameter of the method and is controlled by an error
signal. Error signal has been presented in [117,118] as (46).
By replacing (44) in (43), relation (45) can be concluded.
I dI
∆I I e= +
⎧ ∆ V <− V on the right of MPP V dV (46)
⎪ ∆I I
= −V at MPP According to (45), "e" in relation (46) should be zero at MPP but
⎨∆ V
⎪ ∆ I >− I on the left of MPP actually it cannot be zero because of oscillation, so in this method an
⎩ ∆V V (45) acceptable error range is defined and the formula (46) can be expressed
I ∆I as (47).
is called as instantaneous conductance and as incremental
V ∆V
conductance. Fig. 21 illustrates how this method tracks maximum I dI
power point by comparing instantaneous conductance with incremental e = + <E
V dV (47)
conductance. It should be noticed that big size of incremental con-
ductance will improve speed of method but causes to decrease accuracy Where E is maximum value of acceptable error and tracking style is

Fig. 21. Incremental method flowchart.

13
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

similar to INC method.

3.10.1.2. Variable step size INC. The step size in the INC MPPT method
is generally fixed. If this size is large, the method will be fast with large
oscillations and low efficiency and If step size is small, the result will be
reverse. For tradeoff between speed and oscillation, variable step size
INC method has been presented. This method regulates the step size of
changing in duty cycle of converter in order to track the PV array Fig. 22. Equivalent circuit of PV cell.
maximum power point, so it has higher efficiency, speed and accuracy
rather than fixed step size INC method and also is easy to implement perturbation is applied to array and then average ripple of array's
[119]. For tracking MPP, voltage, current and power of cell are output power and voltage are measured by using of series filters and
measured (V(0),I(0),P(0)) and initial duty cycle (D(0)) is calculated array conductance is calculated. After that traditional INC method can
and applied. If new operating point satisfies the MPPT condition in be used to track MPP [122].
(45), the operating point is considered as MPP. If MPPT condition is not ∞
satisfied, then voltage, current and power of current point are measured iP (t ) = I pDC + ∑ an cos(nwt ) + bn sin (nwt )
again (V(1),I(1),P(1)) and new duty cycle (D(1)) is calculated. All of 1 (52)
these stages will be repeated until MPP is achieved by (45). Duty cycle ∞
of converter can be calculated by (48) where D(K), V(K) and I(K) are vP (t ) = vpDC + ∑ cn cos(nwt ) + dn sin (nwt )
respectively duty cycle, output voltage and current at K-time. 1 (53)
dP ∆P ∞
D (k ) = D (K −1) ± N ≅ D (K −1) ± N P = vpDC . I pDC +0.5 ∑ an cn + bn dn
dV ∆V
1 (54)
P (K ) − P (K −1)
= D (K −1) ± N The admittance of array is Yp = gp + jcpω0, so the lost power in
V (K ) − V (K −1) (48)
array can be calculated by (55).
dP
N is called as scaling factor and N dV as change in duty cycle (step ∞ ∞
size) which should be less than largest step size : ⎛ ⎞
Pgp = 0.5 ∑ an cn + bn dn = g ⎜0.5 ∑ cn2 + dn2⎟
dP
1 p⎝ 1 ⎠ (55)
N <∆Dmax ∞ 2
dV (49) Where (0.5 ∑1 cn2 + dn2) is equal to V0 .
Pgp
gp =
3.10.1.3. Improved variable step size INC method. This is a combination V02 (56)
of fixed and variable step size INC method which uses a threshold
∆C Pgp is the average ripple of power, V0 is the quantity of the voltage
function ( ∆ I ) to switch between fixed and variable step size modes
[120]. ripple and gp is the array conductance [123].
Main advantages: online, medium speed, very high accuracy, good
dP efficiency, adaptability to fast changing conditions.
C = Pn
dI (50) Main drawbacks: parasitic capacitance is notable just in large PV
arrays, parasitic capacitance is impressionable from converter's capa-
Where P and I are power and current of array and n is an index.
citor.
∆C
⎧ ∆ I ≥0 Fixed step size mode (on left of MPP )
⎪∆C 3.10.3. Sliding mode method
⎪ ∆ I <0 variable step size mode (on leftof MPP ) Sliding mode method is a member of conductance based methods
⎨ ∆ C > 0 variable step size mode (on right of MPP ) and its story is so similar to INC method. In the method, at the MPP, "k"
⎪ ∆I
⎪ ∆ C ≤0 Fixed step size mode (on right of MPP ) will be zero where it is defined as the sum of instantaneous conductance
⎩ ∆I (51) ( I ) and incremental conductance ( ∆ I ) according to (45). Control signal
V ∆V
for switching in the converter which is called "tr" here, can be defined as
(57).
3.10.1.4. Power-increment-aided incremental conductance (PI-INC)
method. Ref. [121] presents a method which tracks MPPT by using of 0 k ≥0
VFCD or CFVD control mode in two-stage process by considering a tr = ⎧
⎨
⎩ 1 k<0 (57)
threshold-tracking zone. Where VFCD is variable-frequency with
constant duty and CFVD is constant-frequency with variable duty. When "tr" is zero converter's switch is open and when equals to "1",
The main idea of the method is calculation of power and making sure switch is close [124].
that it is in the specific range. In first stage, voltage and current of cell Main advantages: online, fast, medium accuracy, very good effi-
are measured and then two boundary points are defined on the V-I cency.
curve in neighborhood of MPP to create a zone called as CTZ, these two Main drawbacks: dependence of speed and accuracy on increment
boundary points are ρ1 and ρ2. The powers of boundary points, Pρ1 and size, expensive, complexity.
Pρ2, are calculated and set as bounds of PTZ area on the P-V curve.
Power of operation point is checked to be in [Pρ1, Pρ2]. 3.11. Ripple correction control method (R.C.C.)

3.10.2. Parasitic capacitance method This method is based on utilizing ripple of current and voltage for
This is a developed style of incremental conduction method which tracking MPP. Connecting solar array to converter causes to appear
uses Fig. 22 as equivalent circuit of PV cell instead of Fig. 6-a. As shown ripple in voltage, current and power of PV array and the reason is
in Fig. 22, there is a parallel capacitance in equivalent circuit of cell switching in converter [125,126]. The main difference between this
which INC method does not care it and this is the only difference be- method and perturbation based methods is how the ripple is generated.
tween INC and parastic capacitance method. So like INC method a In RCC method ripple is generated naturally because of switching

14
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

without needing for additional components while in perturbation based

methods an artificial perturbation is needed. By measuring value of
current (or voltage) ripple and power ripple and using of (58), the
operation point can be located and then by increasing or decreasing of
current (or voltage), MPP can be achieved [127].
dI dP
⎧ dt dt > 0 i < iMPP
⎪ dI dP
= 0 i = iMPP
⎨ dt dt
⎪ dI dP < 0 i > iMPP
⎩ dt dt (58)
Fig. 23. Topology of DC Link Capacitor Droop Control method.
It should be noticed that voltage (v) can be replaced with current (i)
in (58).
power of PV array will rise and this can be done as long as consumed
A suitable way for adjusting current is using of boost converter
power by the inverter is not more than maximum power of PV array. In
because in this condition, induction current equals to solar array cur-
this order a point will be met that after it increasement of current causes
rent. Duty cycle of this converter is as same as (59).
drooping in the voltage of DC link, this met point is MPP [133].
Main advantages: fast speed, high accuracy, very good efficiency,
D=k ∫ dI dP
dt dt
dt ≈ k ∫ Php Ihp dt
(59) suitable for partial shading condition.
Main drawbacks: complexity and being expensive.
Where k is a constant, Php and Ihp are first-order quantities after filtering
by a HP filter [128].
3.14. Temperature based method
Main advantages: high speed, high accuracy, real MPPT, very good
efficiency.
This is so similar to constant voltage method and even can be
Main drawbacks: not GMMPT, expensive, relative complexity.
considered as modified style of it. Ref. [134] expresses a relation be-
tween VMPP and temperature of cell's surface (T) as (60).
3.12. Current sweep method
VMPP (T ) = VMPP (Tref ) + uVMPPT (T − Tref ) (65)
This method has very simple idea. It employs a waveform for
Where VMPP(T) is voltage at MPP in T temperature (desired tempera-
sweeping the current of cell in order to approach the V-I characteristic
ture), VMPP(Tref) is voltage at MPP in reference temperature and uVMPP
of it and these steps will be repeated periodically to update the char-
is temperature coefficient of VMPP. the parameters like VMPP(Tref), uVMPP
acteristic, because the ambient conditions may change. After that V-I
and Tref are extracted from sheets and T is measured by sensor then
characteristic is obtained VMPP can be detected and in the result MPP
VMPP(T) is calculated and set as reference voltage. Controller compares
will be achieved [129,130].
measured voltage with reference value and applies error to converter
i (t ) = f (t ) (60) which causes changes in duty cycle of converter. All of these steps are
recurred at every fixed time to update T and Vmppt(T).
At MPP dP/dt is zero so :
Main advantages: simplicity, ability for track MPP in changing
dP (t ) dV (t ) df (t ) conditions, high accuracy, medium speed.
= 0→f (t ) +V =0
dt dt dt (61) Main drawbacks: low efficiency and need for priori knowledge
df about cell.
For simplification, sweep waveform was defined as f (t ) = k dt or
t
f (t ) = Ae k . where k and A are constants, so relation (61) can be written 3.15. β method
as (62).
This is a method based on approximation which tracks β instead of
⎛k dV (t ) + V ⎞ df (t ) = 0
MPP and β is a variable that estimates MPP.
⎝ dt ⎠ dt (62)
By assuming that df (t )
is not zero, (63) can be concluded. ipv ⎞
dt β = ln ⎜⎛ ⎟ − cVpv
V
⎝ pv ⎠ (66)
dP dV (t )
=k +V
di dt (63) error = βg − β (67)
Main advantages: one variable, online.
Dnew = Dold + error . k (68)
Main drawbacks: slow, not so accurate, low efficiency, expensive
q
and complex. Where iPV and vPV are output current and voltage of cell, is a
c= η . K . T . Ns
coefficient that composed of electron charge (q = 1.60217662 ×
3.13. DC link capacitor droop control method 10−19 C), ideal factor of the junction (η), Boltzmann constant (K =
1.38064852 × 10−23 m2 kg s−2 K−1), temperature (T) and number of
This method has been presented for PV systems which are con- series cells (Ns). D is duty cycle, βg is the value of β in the most common
nected to ac system in parallel. In these structures, PV array is con- temperature and k is a constant. If temperature change can be limited in
nected to a converter and this converter is connected to an inverter by a a specific range, the value of β will change in a certain range which it is
DC link as shown in Fig. 23 [131,132]. Duty cycle of converter can be defined as (β min, β max). So by calculating β by (66) and updating its
calculated by (64). value periodically it will be possible to check if β is within fixed rang or
not. If β is lower than βmin or higher than βmax, an error will be
V
D = 1− generated by (67) to adjust the duty cycle of converter in order to re-
V link (64)
turn of β into defined range. After making sure that β is in the accep-
Where V and Vlink are PV array's voltage and DC link voltage. In the table range, the other methods like hill-climbing can be employed to
event that Vlink is kept in a constant value and inverter's input current is track exact MPP. Fig. 24 illustrates flowchart of this method [135].
increased, out coming power of converter and in result out coming Main advantages: good speed, good accuracy, good efficiency.

15
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

Fig. 24. Flowchart of beta method.

Main drawbacks: not true MPPT (if this method doesn’t utilize other
exact methods), need for two types of sensors (voltage and current
sensor).

3.16. One cycle control method (O.C.C.)

The grid connected PV systems usually have two-phase process. In

the fist step a MPPT method is employed by using of DC/DC converter Fig. 25. A system for using by state space method.
and in the second step an inverter makes AC output from DC input
[136]. In OCC method a single stage inverter is used for getting to both and buck converter is between PV array and load. If it is desired to use
of these goals. Ref. [137–139] have presented configuration of inverter an energy storage, it can be located between PV array and converter
for this purpose. and also buck converter can be replaced with other type converter and
Main advantages : less complexity, inexpensive, high speed and if the system is connected to ac grid, it is necesssary to use an inverter
accuracy, good efficiency. next to the converter or instead of it. A general relation for dynamic
Main drawbacks: not true MPPT mode of system can be expressed as (69) [140].

Ẋ = AX (t ) + B (t ) u (t ) + Dε (t ) (69)
3.17. State space method
where x, t, u and ε are respectively considered as state variable vector,
This method utilizes state space equals of the system for tracking time, control input (duty cycle) and disturbance. By solving the equals
MPP. Let assume that Fig. 25 shows the system overview. Cs is a filter of system, the following relations can be presented.

16
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

1 1
x1̇ = vs (t ). u − x2
L L (70)

1 1
x2̇ = x1 − ε (t )
C C (71)
So the relation (69) can be written as (72).
1
⎡0 − L⎤ ⎡ vs (t ) ⎤ u + ⎡ 0 1 ⎤ εy = [0 1] x
1
ẋ = ⎢ 1 ⎥x + ⎢ L
0 ⎥ ⎥ ⎢− ⎥
⎢C
⎣ ⎦ ⎣ 0 ⎦ ⎣ C⎦ (72)
Where u=kT x +mT g1 + nε (t ) , Vs is output voltage of PV array. k, m and n
are controller parameters. kT = [k1 k2], m = [m1 m2 m3] and g1 is Fig. 26. Law table for MPP.
calculated from (73).
∂v
⎡ g (t ) = vo + io ( ∂ioo ) ⎤
⎢ ⎥
⎢
⎢
(
d
( ))
∂vo
g1 = ⎢ g ̇ (t ) = dt vo + io ∂io
⎥
⎥
⎥
⎣
(
dt
2
( ))
⎢ g¨ (t ) = d 2 vo + io ∂vo
∂io
⎥
⎦ (73)
It should be noticed that refs. [141,142] have been presented this
method with more details.
Main advantages: good efficiency, medium speed, effective in par-
tial shaded, suitable in conditions changing.
Fig. 27. Artificial neural network block diagram.
Main drawbacks: being expensive, complexity, need for two types of
sensors.

3.18. Intelligent methods

These are methods which are based on artificial intelligence or

nature behavior so utilizing these methods requires to control knowl-
edge.
Main advantages: good and adjustable speed, good and adjustable
accuracy, no overshoot, less fluctuations.
Main drawbacks: need for control knowledge, complexity, being
expensive.

3.18.1. Fuzzy logic based methods

Fuzzy logic is used to make numerical information from human
knowledge and linguistic variables or to use them for processing nu-
merical data. In the first stage numerical data are converted to linguistic
phrases or membership functions by fuzzy maker, this stage is called as Fig. 28. Depiction of the velocity and position updates in PSO.
fuzzification. In the next stage IF-THEN laws are made for controlling
system according to relations among variables and/or human knowl-
edge about the problem. In the last step results will be converted from
linguistic phrases or fuzzy functions to numerical data. This method has
been explained and implement in many papers like ref. [143–149]. One
of the most common algorithms for MPPT by fuzzy logic has been ex-
plained in [150] :
dP dP
e (k ) = (k ) − (k −1)
dV dV (74)
Fig. 29. Individual structure in genetic algorithm.
ce (k ) = e (k ) − e (k −1) (75)
Where e (k) is error of operation point (deviation from MPP) in instant k Main advantages : high and adjustable speed, very high and ad-
and ce(k) is change of error in instant k. At MPP e (k) is zero so ce(k) justable accuracy, no overshoot, less fluctuations in conditions change,
shows the direction of error. Number of laws depends on desired ac- ability to work with imprecise inputs, very good efficiency
curacy and defined functions. By five functions defined for "e" and "ce" (Figs. 27–29).
twenty five laws (5*5) will be required as in Fig. 26. Main drawbacks : complexity, being expensive, offline.
Where NB is negative big, NS is negative small, ZO is zero, PS is
positive small and PB is positive big. For example out coming law of the
first row and first column means : if error is negative big and change of 3.18.2. Artificial neural network (A.N.N.)
error is negative big then duty cycle change should be zero. Other One of the newest and the most effective methods for tracking MPP
common fuzzy algorithms for tracking MPP has been presented in is artificial neural network. In this method, input variables and goals
[151]. are shown as nods and some links connect nods to each other while any

17
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

link has a special weight. For tracking MPP by this method, effective Gbest will be accepted as the answer. This method has been explained
factors on power of PV cell like irradiance, temperature and voltage are in [172–182]. If Vik and Xik are speed and location of particle "i" in
set as input variables and output can be set as VMPP or any other factor instant "k" respectively, they can be expressed as (79) and (80).
related to MPP like duty cycle of converter. The weights of links are
Vik +1 = wVik + c1 r1 (Pbesti − Xik ) + c2 r2 (Gbest − Xik ) (79)
extracted from relations among system parameters and /or experi-
mental knowledge. This method has been studied in [152–168]. Xik +1 = Xik + Vik +1 (80)
Main advantages: high speed, high accuracy, no overshoot, less
fluctuations in conditions change, good efficiency. Where c1 and c2 are two positive constants, r1 and r2 are two random
Main drawbacks: complexity, being expensive and offline. numbers within [0,1], w is factor of inertia, Pbesti is the best position of
particle "i", Gbest is the best position of whole particles and k is iteration
3.18.3. firefly algorithm index.
This is so powerful algorithm for optimization which is inspired by Main advantages: high speed, high accuracy, good efficiency.
fireflies and has been introduced in [169,170] by Xin-She Yang in 2008. Main drawbacks: complexity and being expensive.
There are three suppositions in the algorithm.
3.18.5. Genetic algorithm
1) All of fireflies are same in sex so sexuality has no role in attraction of Genetic algorithm is an efficient method for optimization which is
one firefly to one other. inspired by process of nature evolution and is composed of three basic
2) The attraction between every two fireflies depends on their relative operators as "selection", "crossover" and "mutation". The first one re-
brightness in order to firefly with less bright will attract to firefly presents theory of natural selection and expresses that always the best
which is seen brightest. If the brightness of fireflies are same, every generation is chosen to survive. Crossover operator plays the role of
one move randomly. intercross to generate new generation which are better and more dur-
3) Whatever the location of a firefly is closer to the objective function, able. The third operator, mutation, makes variety in genes' character-
it will be brighter. istics and possibility of global optimization [183,184]. This algorithm
starts the search of optimized point by some initial genes (first gen-
in the first stage, some fireflies are randomly located in the answer eration) then the fitness of them are calculated and the fittest of them
space and their fitness are calculated by using of their location and are selected for mating (crossover). After that new genes (second gen-
objective function, then distance and attraction between every two eration) are generated, the fitness of them will be calculated and the
fireflies are calculated by (76) and (77) respectively. fittest will be picked up for crossover and all of these stages will be
repeated until stop condition is met. At last stage, the fittest gene is
dij = Xi − Xj (76)
considered as the answer. It is clear that for generating random number
βij (dij ) = β0i e−γ (dij)
n for using as mutation, random signal is required. So genetic algorithm
(77)
needs to be implemented on the digital system. For MPPT, the fitness
i and j are counters for fireflies such that i,j = 1,2,… and i ≠ j, Xi and value is the power level [185].
Xj are location of i and j fireflies. dij is distance between i and j, βij (dij ) is
x n = (a. x n −1 + c ). (mod ). m (81)
the degree of attractiveness between i and j and β0i is the initial at-
tractiveness of i firefly which is chosen randomly. γ is coefficients Where xn is current random number and "mode" is modulus operator.
which is bigger than 0 and n is a constant and greater than one. After Main advantages: high speed, high accuracy and good efficiency.
calculation of distance and attractiveness, every firefly (for example "a") Main drawbacks: complexity and being expensive.
is compared with the rest of fireflies and whom have fitness more than
"a" are selected, then firefly with highest attractiveness (highest βaj) is 3.19. Hybrid methods
chosen among them and firefly "a" starts to go toward it. For example if
"a" goes toward "b", the new position of firefly "a" can be presented by Any method has its own advantages and drawbacks so by correct
(78) [171]. combining of conventional methods can make a new method which has
more benefits than usual methods. For example P & O is one of the most
Xat+1 = Xat + βab (dab)*dab + α (rand−0.5) (78)
common methods but its speed depends on perturbation size. If this size
Where "α" is random number in [0,1] which is called as movement is small, method will have so good accuracy but low speed. For solving
factor and "rand" is another random number in that interval. For this problem, it is possible to combine this method with artificial neural
tracking MPP by this method, the position of firefly is considered as network [186] or open circuit voltage. The other example can be esti-
duty cycle and the brightness is looked as output power of cell. mation and revision method [187] or combination of ANN and genetic
Main advantages: high speed, very high accuracy and very good algorithm [188] or combination of fuzzy logic and genetic algorithm
efficiency. [189]. Also ref. [190] combines genetic algorithm and P & O method.
Main drawbacks: being expensive and complex.
4. Comparison of different MPPT methods
3.18.4. Particle swarm optimization (PSO) method
This method considers a set of hypothetical particles which are Main advantages and disadvantages of each method expressed in
moving in search area with adjusted speed. Any particle has a limited the part (3) which can be a great help to choose a suitable method,
memory which can save its own and its neighbor particles' the best nevertheless in this part the main factors for choosing a method are
positions. At first step, some particles are distributed in search area discussed.
randomly and initial position of each particle is saved as best position of
it (Pbest) and also the best position among the all of them is saved as 4.1. Performance
the global best position (Gbest). At next step, a velocity vector is ap-
plied to every particle and new positions of the particles are calculated This is one of the most important factors in choosing a method and it
and compared with saved positions as Pbest and Gbest to update them. is clear why. three parameters which make this factor are: speed, ac-
All of these steps will repeat until stop condition is met and finally curacy and efficiency of energy. In the following, these parameters are

18
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

evaluated for each of the explained methods in part (3). Table 1

Curve fitting method: this is a low speed method mainly due to the Comparison of different P & O methods.
need for large number of calculations to track MPP. This method just
Traditional PI type M type Two-section
approximates the MPP and is not a "true method" and also depends on P&O type
ambient conditions extremely so accuracy and efficiency of this method
are low. Rise time (sec) 1.3002 0.1004 0.7007 1.3002
Steady state time (sec) 1.3023 0.1007 0.8001 1.3023
Look up table: this is another low speed method and its reason is
Steady state average 238.73 240.45 240.60 239.64
that this method should measure the conditions and then search among power (W)
data base to find a matched or similar data. The accuracy and efficiency Steady state tracking 99.15 99.87 99.93 99.53
of this method are low because of reasons as same as that of curve accuracy (%)
fitting method. Tracking energy loss 149.85 11.87 83.13 148.10
(J)
Power matching: this is so simple method which doesn't need any
Average tracking 49.95 3.96 27.71 49.37
converter or other additional equipment even can be used without any power loss(W)
sensor so it is clear this method is inexpensive but accuracy and effi-
ciency is low.
Numerical methods: these offline methods have a good speed gen-
erally. Newton- Raphson, Secant method, linearization and central
point method are fast because of their algorithms that converge rapidly.
Steepest descent and regula falsi have medium speed but bisection is
quite slow. All of these methods are high precision thanks to their ac-
curate mathematical theorems and algorithms.
FSCC and FOCV methods: these two methods are so similar to each
so they will have the same result in tracking MPP. These methods have
medium speed in tracking desired point and it can be because they need
to calculate current and voltage factors before changing duty cycle of
converter and even in some cases, system should measure short circuit
current and open circuit voltage to improve accuracy. It is clear that if
current and voltage factors are assumed constant the speed can become
a little high and also the methods will be simpler but that accuracy will
decrease. The accuracy and efficiency of these methods are low and the
reason is estimation. These methods just estimate MPP and are not "true
methods". The other reason for low efficiency is that PV array is dis-
connected from system when Isc and Voc are measured and this means
that generated power will lose. Fig. 30. Comparison of some common methods' tracking factor.
Constant voltage method: speed does not mean in this case because
as can identify from the name, voltage is constant always and operation Table 2
point is constant so according to amount has been set as constant value Comparison of P & O, FL with PSO in shading conditions.
for voltage, efficiency and accuracy can be different but efficiency and
accuracy will be low even in the best conditions. Test NO Tracking Power (W) time Max. Power Tracking
methods (sec) from P-V efficiency (%)
Differential based methods: the main methods of this group are
curve
power variation with time and with voltage (dP/dt and dP/dV). Both of
these methods have high accuracy and good efficiency thanks to their 1 P&O 31.4886 12.7 41.59 90.14
exact algorithms but the method "dP/dt" needs for too many calcula- PSO 41.4538 13.6 99.67
FUZZY 41.576 2.1 99.97
tions which causes this method be slow for tracking MPP while "dP/dV"
2 P&O 43.6957 8.8 66.56 65.65
method requires less calculation so its speed is higher than "dP/dt" PSO 66.4869 7.8 99.98
method. FUZZY 66.5577 1.96 99.99
Perturbation based methods: this family of MPPT methods includes
P & O, MP & O, hill climbing, forced oscillation, EPP and three point
weight methods. All of these methods detect MPP by applying pertur- Table 3
Comparison of GA with PSO.
bation to one or two control parameters and comparing the result with
previous point. The perturbation size is so important in speed and ac- Irradiation variations Efficiency
curacy of these methods. Big size means less measurements and cal-
culation and also higher speed but will pull accuracy down while small GA PSO
size means lower speed and higher accuracy, so a tradeoff between
G1 = 1000 W/m2 and G2 = 600 W/m2 0.956 0.975
speed and accuracy is necessary which causes medium speed for these G1 = 600 W/m2 and G2 = 600 W/m2 0.966 0.994
methods. Modified P & O and three point weight methods are similar to G1 = 800 W/m2 and G2 = 600 W/m2 0.946 0.985
P & O but their algorithms are faster and accordingly these two methods G1 = 1100 W/m2 and G2 = 600 W/m2 0.944 0.97
go into fast class. In accuracy, P & O method has a medium algorithm
while other methods of this family which actually are advanced forms
of P & O have high accuracy algorithms. Table 1 compares traditional Conduction based methods: the main members of this group are INC
P & O with modified P & O methods including: PI type P & O, M type method, parasitic capacitance method and sliding method. fundamental
P & O and two-section type P & O by a simulation [191]. concept of INC method is so similar to perturbation methods but in this

19
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

Table 4
Comparison of GA and PSO with firefly algorithm.

Functions Algorithms GA PSO FIREFLY

Michalewiez's (d = 16) 95% 98% 99%

Rosenbrock's (d = 16) 90% 98% 99%
De Jong's (d = 256) 100% 100% 100%
Schwefel's (d = 128) 95% 97% 100%
Ackley's (d = 128) 90% 92% 100%
Rastrigin's 77% 90% 100%
Easom's 92% 90% 100%
Griewank's 90% 92% 100%
Shubert's (18minima) 89% 92% 100%
Yang's (d = 16) 83% 90% 100%

method perturbation is not artificial which causes fast speed for OCC method: using of one cycle inverter and just one type sensor
tracking MPP. High accuracy and good efficiency are other character- (current sensor) makes this method inexpensive with high speed and
istics of this method. Parasitic capacitance method is a comprehensive accuracy along with good accuracy.
form of INC method so the specifications of this method are similar to State space method: good efficiency and high accuracy in this
INC's. Parasitic capacitance method has very high accuracy and good method are appearing as in any other mathematical based methods.
efficiency in its own list of advantage but taking account of parallel Value of Ẋ is depend on time so this value changes continuously which
capacitance's effect causes reduction of speed and makes medium speed makes this method medium speed.
for this method. Sliding mode uses concept of INC method to control Intelligent methods : speed and accuracy of these methods relate to
switching of converter. Fast speed, medium accuracy and very good different parameters for example in fuzzy logic method increasement of
efficiency are properties of this method. the number of laws means increasing the accuracy but it will need to
RCC method: this method utilizes nature oscillations which are more time to track MPP and speed will decrease. In PSO, firefly and
generated because of switching to locate MPP and in result this method other particle based methods the number of the particles, search area
has high accuracy and good efficiency, like hill climbing method but and etc define accuracy and speed. Intelligent methods have fast speed,
they are different in speed. Hill climbing was a medium speed method high accuracy and good efficiency in general.
while RCC is fast and expensive. Comparing among intelligent methods is really difficult because
Current sweep method: the speed is low because this method needs their speed and accuracy will be change by changing the laws, number
to sweep P-V characteristic in fixed times for locating MPP. Ambient of functions or partials or weight and etc but some papers have done it.
conditions like irradiance and temperature may change rapidly which The most common intelligent methods for tracking MPP are fuzzy logic
cause changing of P-V characteristic and MPP location but until next control and PSO. Ref. [169] compares P & O and fuzzy logic with PSO in
sweeping the system cannot sense this change and some power will be two different shading conditions. The result of this comparison is shown
lost, so the accuracy of this method is medium while its efficiency is in Table 2.
low. As shown in the last column of Table 2, PSO and FL methods have so
DC link capacitance droop: this method utilizes a simple and routine good accuracy but FL accuracy is a little more than PSO. Ref. [192]
algorithm which helps it have high speed. Other benefits of this algo- compared genetic algorithm with PSO by simulation in Matlab/Sim-
rithm are high accuracy and efficiency. power system. Table 3 shows the result, also comparison of three in-
Temperature method: temperature of cell should be measured and telligent methods: PSO, genetic algorithm and firefly algorithm has
VMPP is calculated and then like constant voltage method, calculated been done in [168] which its result is shown in Table 4.
value is set as reference. In the next step present value of voltage is
measured and compared with reference value and error goes to con- 4.2. Cost
troller to adjust duty cycle of converter and measuring and comparing
are repeated until MPP achieved. According to these steps it is clear that One of the important factors and maybe the most important in
this method needs long time for tracking MPPT and is not a high speed choosing a method is cost. So tradeoff between efficiency and cost is
method but the accuracy of this algorithm is high because there is a necessary to get desired goal. Two parameters which have the most
close relation between temperature and MPP and this relation has been effect on cost of a method are sensors and microcontroller. Curve fitting
considered in this method. It should be noticed that temperature can and look-up table methods don't need to high computation micro-
change rapidly and this will cause changing in MPP location but this controller and power matching method doesn't need even sensor so
method doesn’t sense it until updating measured temperature and this these methods are inexpensive. Almost all of the numerical methods are
can reduce the efficiency. expensive because they should do a large number of calculations to
Beta method: this method approximates MPP by a variable called achieve MPP so they need high computation microcontroller. FSCC,
beta and the approximation is the reason that makes this method high FOCV, CV and load based methods all are inexpensive thanks to their
speed. It is clear that estimation will decrease the accuracy but this simplicity and a few numbers of computation while needing for high
method is often switched to an exact method like INC after that closed computation microcontroller, large amount calculation and needing for
to MPP so it will be true if say this method is an exact method with good two types of sensors make differential methods expensive.
efficiency. Fig. 30 compares some common methods in efficiency Perturbation methods need to have additional components for
(tracking factor) [136]. generating perturbations and these components along with good

20
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

Table 5
Comparison of different MPPT methods.

Method Economic Performance Control circuit

Cost True Speed Accuracy Energy sensors PV array Complexity Parameter On line Partial A/D App.
MPPT tracking dependence tuning shading
factor

Curve fitting INEX NO SLOW LOW LOW V YES SIMPLE YES NO NO DIGITAL S.A
Look- up table INEX YES SLOW LOW LOW V,I YES SIMPLE YES NO NO DIGITAL S.A
Power INEX NO FAST LOW LOW NONE YES SIMPLE NO NO NO ANALOG S.A
matching
Newton- EX YES FAST HIGH GOOD V,I NO SIMPLE YES NO NO DIGITAL S.A
Raphson
Secant EX YES FAST HIGH GOOD V,I NO SIMPLE YES NO NO DIGITAL S.A
Linearizat-ion INEX YES FAST HIGH GOOD IRRAD YES SIMPLE YES NO NO DIGITAL S.A
Steepest EX YES MEDIUM HIGH GOOD V OR I NO MEDIUM NO NO NO DIGITAL S.A
descent
Central point EX YES FAST HIGH GOOD I NO COMPLEX YES NO NO DIGITAL S.A
Bisection INEX YES Slow HIGH GOOD V,I NO MEDIUM YES NO NO DIGITAL S.A
Regula falsi EX YES MEDIUM HIGH GOOD V,I NO MEDIUM YES NO NO DIGITAL S.A
FSCC INEX NO MEDIUM LOW LOW I YES SIMPLE YES NO NO BOTH S.A
FOCV INEX NO MEDIUM LOW LOW V YES SIMPLE YES NO NO BOTH S.A
CV INEX NO N/A LOW LOW NONE YES SIMPLE YES NO NO BOTH BOTH
Load based INEX NO FAST HIGH LOW V OR I NO SIMPLE NO NO NO ANALOG S.A
Differential EX YES SLOW HIGH HIGH V OR I NO COMPLEX YES YES NO DIGITAL S.A
( dP )
dt
Differential EX YES MEDIUM HIGH HIGH V,I NO COMPLEX NO YES NO DIGITAL S.A
( dP )
dV
P&O EX YES MEDIUM MEDIUM GOOD V,I NO SIMPLE NO YES NO DIGITAL BOTH
MP & O EX YES FAST HIGH HIGH V,I NO MEDIUM NO YES YES DIGITAL BOTH
EPP EX YES MEDIUM HIGH HIGH V,I NO COMPLEX NO YES BOTH BOTH
Hill climbing EX YES MEDIUM HIGH GOOD V,I NO SIMPLE NO YES NO DIGITAL S.A
Forced EX YES MEDIUM HIGH GOOD V OR I NO COMPLES YES YES NO ANALOG S.A
oscillation
Three point EX YES FAST HIGH HIGH V,I NO COMPLEX NO YES NO DIGITAL BOTH
weight
INC EX YES FAST HIGH GOOD V,I NO MEDIUM NO YES YES DIGITAL BOTH
Parasitic EX YES MEDIUM VERY HIGH GOOD V,I NO COMPLEX YES YES YES DIGITAL S.A
capaci-
tance
Sliding mode EX YES FAST MEDIUM VERY V OR I NO COMPLEX NO YES YES DIGITAL BOTH
GOOD
RCC EX YES FAST HIGH VERY V OR I NO MEDIUM NO YES YES ANALOG S.A
GOOD
Current sweep EX YES SLOW MEDIUM LOW I YES COMPLEX YES YES NO DIGITAL S.A
DC link cap. EX YES FAST HIGH VERY V NO COMPLEX YES YES YES BOTH BOTH
droop GOOD
Temperature EX YES MEDIUM HIGH LOW V,T YES MEDIUM YES YES NO DIGITAL S.A
based
Beta EX YES FAST HIGH GOOD V,I YES COMPLEX NO NO YES DIGITAL GRID
OCC INEX NO FAST HIGH GOOD I YES MEDIUM YES NO BOTH BOTH
State space EX YES MEDIUM HIGH GOOD V,I YES COMPLEX YES YES DIGITAL BOTH
Fuzzy logic EX YES FAST VERY HIGH VERY V OR I NO COMPLEX YES NO YES DIGITAL BOTH
GOOD
ANN EX YES FAST VERY HIGH GOOD V,I NO COMPLEX YES NO YES DIGITAL GRID
Firefly EX YES FAST VERY HIGH VERY V,I NO COMPLEX NO YES DIGITAL S.A
algorithm GOOD
PSO EX YES FAST VERY HIGH GOOD V,I NO MEDIUM NO YES YES DIGITAL S.A
Genetic EX YES FAST Very HIGH GOOD V,I NO MEDIUM NO YES DIGITAL S.A
algorithm

Where EX = expensive, INEX = inexpensive, IRRAD = irradiance and S.A = stand alone.

microcontroller and two types sensors makes this family expensive. alone, grid connected and hybrid. It is necessary to choose a method
Although some other methods like INC, parasitic capacitance, DC link, which is suitable for required application. Application of each method
sliding method RCC, temperature and beta methods don't need for has been expressed in Table 5.
perturbation generating equipment but they are still expensive.
State space method needs to high level microcontroller in addition
of voltage and current sensors. Intelligent methods have a same story so 4.4. Digital or analog
both of state space and intelligent methods are expensive.
Another factor which can be important is that a method is analog or
digital. This factor's importance depends on required application
4.3. Application
(Table 6).
There are three different type of application for PV arrays: stand

21
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

Table 6 Retrieved February; 2015.

definition of used parameters. [7] Esram T, Chapman PL. Comparison of photovoltaic array maximum power point
tracking techniques. IEEE Trans Energy Convers 2007;22:439–49.
[8] Leedy AW, Garcia KE. Approximation of P-V characteristic curves for use in
Symbol Parameter
maximum power point tracking algorithms. In: 2013 Proceedings of the 45th
Southeastern Symposium on System Theory (SSST); 2013, pp. 88–93.
P Power
[9] KHATIB ATN, MOHAMED A, AMIN N, Sopian K. An efficient maximum power
V Voltage point tracking controller for photovoltaic systems using new boost converter de-
VMPP Voltage at maximum power point sign and improved control algorithm. WSEAS Trans Power Syst 2010;5:53–63.
Voc Open circuit voltage [10] Ibrahim HESA, Houssiny FF, El-Din HMZ, El-Shibini MA. "Microcomputer con-
I Current trolled buck regulator for maximum power point tracker for DC pumping system
IMPP Current at maximum power point operates from photovoltaic system. In: Proceedings of 1999 IEEE International
Isc Short circuit current Fuzzy Systems Conference, vol. 1. FUZZ-IEEE '99; 1999. p. 406–11.
Is Generated current by cell [11] Yang C, Smedley K, Vacher F, Brouwer J. A new maximum power point tracking
Io The reverse saturation current of cell controller for photovoltaic power generation. In: Proceedings of eighteenth annual
RS The series internal resistance of the cell IEEE Applied Power Electronics Conference and Exposition, vol. 1. APEC '03;
K Boltzmann's constant 2003. p. 58–62.
[12] Charfi S, Chaabene M. Acomparative study of MPPT techniques for PV systems. In:
Q Charge of one electron
Proceedings of 2014 5th International Renewable Energy Congress (IREC); 2014.
A Ideality factor for p-n junction of cell
p. 1–6.
T Temperature
[13] Alghuwainem SM. Performance analysis of a PV powered DC motor driving a 3-
Tcell Temperature on cell surface phase self-excited induction generator. IEEE Trans Energy Convers
t time 1996;11:155–61.
G Irradiance value on cell surface [14] Shengyi L, Dougal RA. Dynamic multiphysics model for solar array. IEEE Trans
X A typical variable Energy Convers 2002;17:285–94.
F(x) A typical function [15] Ishaque K, Salam Z, Taheri H, Syafaruddin. Modeling and simulation of photo-
Fs Switching frequency voltaic (PV) system during partial shading based on a two-diode model. Simul
Fm Modulating frequency Model Pract Theory 2011;19:1613–26.
n and i Index of iterate [16] Parker A, Hamblen JO. Optimal value for the Newton-Raphson division algorithm.
ԑ a small positive infinitesimal quantity as acceptable error (tolerance) Inf Process Lett 1992;42:141–4.
ln(x) Tangent function [17] Ng SW, Lee YS. Variable dimension Newton-Raphson method. IEEE Trans Circuits
hn(x) Secant function Syst I: Fundam Theory Appl 2000;47:809–17.
[18] Abbasbandy S. Improving Newton–Raphson method for nonlinear equations by
αk Step size in steepest descent method
modified Adomian decomposition method. Appl Math Comput 2003;145:887–93.
Kε Step size corrector in steepest descent method
[19] Daumas M, Revol N, Kornerup P, Muller J-M. Real numbers and computers
−∇f (x k ) Direction of the descent in steepest descent method choosing starting values for certain Newton–Raphson iterations. Theor. Comput
Ki Current factor Sci 2006;351:101–10.
Kv Voltage factor [20] Ne D, x, x, Tan Y, Moase WH, Manzie C. A unifying approach to extremum seeking:
∆Tr Threshold value Adaptive schemes based on estimation of derivatives. In: Proceedings of the 49th
D Duty cycle of converter IEEE Conference on Decision and Control (CDC); 2010, p. 4625–30.
[21] Ghaffari A, Krsti M, Nešić D. Multivariable Newton-based extremum seeking [8//].
Automatica 2012;48:1759–67.
[22] Yang D, Qi C, Ming Z. International conference on applied physics and industrial
5. Conclusions engineering 2012 photovoltaic module simulink model for a stand-alone PV
system. Phys Procedia 2012;24:94–100.
More than forty one MPPT methods have been expressed and their [23] Zazo H, Leyva R, Castillo Ed. MPPT based on Newton-Like extremum seeking
control. In: Proceedings of 2012 IEEE International Symposium on Industrial
advantages and drawbacks have been discussed. Although there are still Electronics (ISIE); 2012. p. 1040–5.
more methods which they have not been listed in this paper but the [24] Brahmi H, Dhifaoui R. Dynamic characteristics and improved MPPT control of PV
most common methods have been. Here, there are some suggestions generator. Front Energy 2013;7:342–50.
[25] Hosseini SH, Farakhor A, Haghighian SK. Novel algorithm of MPPT for PV array
about application of each method. Methods like curve fitting, look up
based on variable step Newton-Raphson method through model predictive control.
table, FSCC, FOCV, CV and load base method can be suitable for small In: Proceedings of the 13th International Conference on Control, Automation and
scale applications where cost is more important than efficiency and Systems (ICCAS); 2013. p. 1577–82.
[26] Moura SJ, Chang YA. Lyapunov-based switched extremum seeking for photo-
accuracy like domestic applications solar chargers. Differential based
voltaic power maximization. Control Eng Pract 2013;21:971–80.
methods, numerical methods, perturbation based and conductance [27] Khaldi N, Mahmoudi H, Zazi M, Barradi Y. The MPPT control of PV system by
based methods are suitable for commercial and usual applications. State using neural networks based on Newton Raphson method. In: Proceedings of 2014
space and intelligent methods are for applications with high sensitivity International Renewable and Sustainable Energy Conference (IRSEC); 2014, p.
19–24.
where efficiency, accuracy and speed are so important like in space [28] Li H, Peng J, Liu W, Huang Z, Lin K-C. A Newton-based extremum seeking MPPT
applications. It should be noted that these proposals are general so it is method for photovoltaic systems with stochastic perturbations. Int J Photo
necessary to study, compare and tradeoff among important factors be- 2014;2014:13.
[29] Chin VJ, Salam Z, Ishaque K. Cell modelling and model parameters estimation
fore choosing any method. Table 5 shows the main factors of any techniques for photovoltaic simulator application: a review. Appl Energy
method like cost, performance and etc which should be considered in 2015;154:500–19.
method selection. [30] Maharjan S, Peng JCH, Xiao W. Improved deterministic real-time estimation of
Maximum Power Point in photovoltaic power systems. In: Proceedings of 2015
IEEE 8th GCC Conference and Exhibition (GCCCE); 2015. p. 1–6.
References [31] Uoya M, Koizumi H. A calculation method of photovoltaic array's operating point
for MPPT evaluation based on one-dimensional Newton & Raphson method. IEEE
Trans Ind Appl 2015;51:567–75.
[1] Bull SR. Renewable energy today and tomorrow. Proc IEEE 2001;89:1216–26.
[32] Amir A, Amir A, Selvaraj J, Rahim NA. Study of the MPP tracking algorithms:
[2] Matos FB, Camacho JR. A model for semiconductor photovoltaic (PV) solar cells:
focusing the numerical method techniques. Renew Sustain Energy Rev
the physics of the energy conversion, from the solar spectrum to dc electric power.
2016;62:350–71.
In: Proceedings of 2007 international conference on clean electrical power; 2007.
[33] Chun S, Kwasinski A. Analysis of classical root-finding methods applied to digital
p. 352–9.
maximum power point tracking for sustainable photovoltaic energy generation.
[3] Dubey R. Neural network MPPT control scheme with hysteresis current controlled
IEEE Trans Power Electron 2011;26:3730–43.
inverter for photovoltaic system. In: Proceedings of 2014 Recent Advances in
[34] Kaw A, Kalu EE. Numerical methods with applications: Abridged, second ed.:
Engineering and Computational Sciences (RAECS); 2014. p. 1–6.
autarkaw; 2011.
[4] Salas V, Olías E, Barrado A, Lázaro A. Review of the maximum power point
[35] Allen MB, Isaacson EL. Numerical analysis for applied science. New York, United
tracking algorithms for stand-alone photovoltaic systems. Sol Energy Mater Sol
States: John Wiley & Sons; 1998.
Cells 2006;90:1555–78.
[36] Xiao W, Dunford WG, Palmer PR, Capel A. Application of centered differentiation
[5] Recent facts about photovoltaics in Germany, Fraunhofer ISE, 19 May 2015.
and steepest descent to maximum power point tracking. IEEE Trans Ind Electron
Retrieved 3 July 2015.
2007;54:2539–49.
[6] Photovoltaikanlagen: Datenmeldungen sowie EEG-Vergütungssätze. Monthly re-
[37] Wang P, Zhu H, Shen W, Choo FH, Loh PC, Tan KK. A novel approach of
ported new installations of PV systems and current feed-in tariffs (in German).

22
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

maximizing energy harvesting in photovoltaic systems based on bisection search control. Proc Power Convers Con 1997:691–6.
theorem. In: Proceedings of 2010 twenty-fifth annual IEEE Applied Power [68] Wasynezuk O. Dynamic behavior of a class of photovoltaic power systems. IEEE
Electronics Conference and Exposition (APEC); 2010. p. 2143–8. Trans Power Appar Syst 1983;vol. PAS-102:3031–7.
[38] Zhang Q, Hu C, Chen L, Amirahmadi A, Kutkut N, Shen ZJ, et al. A center point [69] Salameh Z, Taylor D. Step-up maximum power point tracker for photovoltaic ar-
iteration MPPT method with application on the frequency-modulated LLC micro- rays. Sol. Energy 1990;44:57–61.
inverter. IEEE Trans Power Electron 2014;29:1262–74. [70] Teulings WJA, Marpinard JC, Capel A, Sullivan DO. "A new maximum power point
[39] Burden LR, Faires JD. Numerical analysis. Boston: PWS-Kent; 1989. tracking system. In: Proceedings of 24th annual IEEE Power Electronics Specialists
[40] Dahlquist G, Björck Å. Numerical methods. New York: Dover Publications; 2003. Conference. PESC '93 Record; 1993, pp. 833–8.
[41] Goyal M. Numerical methods and statistical techniques using 'C'. New Delhi: [71] Slonim MA, Rahovich LM, Maximum power point regulator for 4 kW solar cell
University Science Press; 2010. array connected through invertor to the AC grid. In: Proceedings of the 31st
[42] Ching-Tsai P, Jeng-Yue C, Chin-Peng C, Yi-Shuo H. "A fast maximum power point Intersociety Energy Conversion Engineering Conference, vol. 3. IECEC 96; 1996,
tracker for photovoltaic power systems. In: Proceedings of the 25th Annual pp. 1669–72.
Conference of the IEEE Industrial Electronics Society, vol. 1. IECON '99; 1999. p. [72] Al-Amoudi A, Zhang L. Optimal control of a grid-connected PV system for max-
390–93. imum power point tracking and unity power factor. In: Proceedings of the Seventh
[43] Noguchi T, Togashi S, Nakamoto R. Short-current pulse-based adaptive maximum- International Conference on Power Electronics and Variable Speed
power-point tracking for a photovoltaic power generation system. Electr Eng Jpn Drives; (Conference Publ. No. 456); 1998. p. 80–5.
2002;139:65–72. [73] Kasa N, Lida T, Iwamoto H. Maximum power point tracking with capacitor iden-
[44] Moldovan L, Kumar KK, Bhaskar R, Koti H. The 7th International Conference tifier for photovoltaic power system. In: IEE proceedings - electric power appli-
Interdisciplinarity in Engineering, INTER-ENG 2013, 10-11 October 2013, Petru cations, vol. 147; 2000, pp. 497–502.
Maior University of Tirgu Mures, RomaniaImplementation of MPPT algorithm for [74] Zhang L, Al-Amoudi A, Yunfei B. Real-time maximum power point tracking for
solar photovoltaic cell by comparing short-circuit method and incremental con- grid-connected photovoltaic systems. In: Proceedings of the eighth international
ductance method. Procedia Technol 2014;12:705–15. conference on power electronics and variable speed drives (IEE Conference Publ.
[45] Sher HA, Murtaza AF, Noman A, Addoweesh KE, Chiaberge M. An intelligent No. 475), 2000, pp. 124–9.
control strategy of fractional short circuit current maximum power point tracking [75] Chomsuwan K, Prisuwanna P, Monyakul V. Photovoltaic grid-connected inverter
technique for photovoltaic applications. J Renew Sustain Energy 2015;7:013114. using two-switch buck-boost converter. In: Proceedings of photovoltaic specialists
[46] Masoum MAS, Dehbonei H, Fuchs EF. Theoretical and experimental analyses of conference, 2002. Conference record of the twenty-ninth IEEE; 2002. p. 1527–30.
photovoltaic systems with voltage and current-based maximum power-point [76] Mao-Lin C, Chih-Chiang H, Jong-Rong L. Direct power control for distributed PV
tracking. IEEE Trans Energy Convers 2002;17:514–22. power system. In: Proceedings of the Power Conversion Conference, vol. 1. PCC-
[47] Andersen M, Alvsten B. 200 W low cost module integrated utility interface for Osaka 2002; 2002. p. 311–5.
modular photovoltaic energy systems. In: Proceedings of the 21st International [77] Ying-Tung H, China-Hong C. Maximum power tracking for photovoltaic power
Conference on Industrial Electronics, Control, and Instrumentation, vol. 1. IEEE system. In: Proceedings of the 37th IAS annual meeting. Conference record of the
IECON; 1995. p. 572–7. industry applications conference, vol. 2; 2002. p. 1035–40.
[48] Ahmad J. A fractional open circuit voltage based maximum power point tracker [78] Youngseok J, Gwonjong Y, Jaeho C, Juyeop C, "High-frequency DC link inverter
for photovoltaic arrays. In: 2010 Proceedings of the 2nd International Conference for grid-connected photovoltaic system. In: Proceedings of photovoltaic specialists
on Software Technology and Engineering (ICSTE); 2010. p. V1-247-V1-250. conference. conference record of the twenty-ninth IEEE; 2002. p. 1410–3.
[49] Raveendhra D, Kumar B, Mishra D, Mankotia M. Design of FPGA based open cir- [79] Chung HSH, Tse KK, Hui SYR, Mok CM, Ho MT. A novel maximum power point
cuit voltage MPPT charge controller for solar PV system. In: Proceedings of 2013 tracking technique for solar panels using a SEPIC or Cuk converter. IEEE Trans
International Conference on Circuits, Power and Computing Technologies Power Electron 2003;18:717–24.
(ICCPCT); 2013. p. 523–7. [80] Youssef EB, Stephane P, Bruno E, Corinne A. New P & O MPPT algorithm for FPGA
[50] Baimel D, Shkoury R, Elbaz L, Tapuchi S, Baimel N. Novel optimized method for implementation. In: IECON 2010 - Proceedings of the 36th annual conference on
maximum power point tracking in PV systems using fractional open circuit voltage IEEE industrial electronics society; 2010. p. 2868–73.
technique. In: Proceedings of 2016 international Symposium on Power Electronics, [81] Bianconi E, Calvente J, Giral R, Mamarelis E, Petrone G, Ramos-Paja CA, et al.
Electrical Drives, Automation and Motion (SPEEDAM); 2016. p. 889–4. Perturb and Observe MPPT algorithm with a current controller based on the
[51] Schoeman JJ, Wyk JDv. A simplified maximal power controller for terrestrial sliding mode. Int J Electr Power Energy Syst 2013;44:346–56.
photovoltaic panel arrays. In: Proceedings of power electronics specialists con- [82] Selmi T, Abdul-Niby M, Devis L, Davis A. P & O MPPT implementation using
ference. IEEE; 1982. p. 361–367. MATLAB/Simulink. In: Proceedings of the ninth international conference on
[52] Ela MAE, Roger J. Optimization of the function of a photovoltaic array using a Ecological Vehicles and Renewable Energies (EVER); 2014. p. 1–4.
feedback control system. Sol Cells 1984;13:107–19. [83] Hussein KH, Muta I, Hoshino T, Osakada M. Maximum photovoltaic power
[53] Hadji S, Gaubert JP, Krim F. Maximum Power Point Tracking (MPPT) for tracking: an algorithm for rapidly changing atmospheric conditions. In: IEE pro-
Photovoltaic systems using open circuit voltage and short circuit current. In: ceedings - generation, transmission and distribution, vol. 142; 1995. p. 59–64.
Proceedings of the 3rd international conference on systems and control; 2013. p. [84] Yan Hong L, Hamill DC. Simple maximum power point tracker for photovoltaic
87–92. arrays. Electron Lett 2000;36:997–9.
[54] Reza Reisi A, Hassan Moradi M, Jamasb S. Classification and comparison of [85] Femia N, Petrone G, Spagnuolo G, Vitelli M. "Increasing the efficiency of P & O
maximum power point tracking techniques for photovoltaic system: a review. MPPT by converter dynamic matching. In: Proceedings of 2004 IEEE international
Renew Sustain Energy Rev 2013;19:433–43. symposium on industrial electronics, vol. 2; 2004. p. 1017–1021.
[55] Hart GW, Branz HM, Cox CH. Experimental tests of open-loop maximum-power- [86] Yun Tiam T, Kirschen DS, Jenkins N. A model of PV generation suitable for sta-
point tracking techniques for photovoltaic arrays. Sol Cells 1984;13:185–95. bility analysis. IEEE Trans Energy Convers 2004;19:748–55.
[56] Maheshappa HD, Nagaraju J, Murthy MVK. An improved maximum power point [87] Abu T, Asghar MSJ. Development of microcontroller-based maximum power point
tracker using a step-up converter with current locked loop. Renew. Energy tracker for a photovoltaic panel. In: Proceedings of 2006 IEEE power india con-
1998;13:195–201. ference; 2006, p. 5 pp.
[57] Lopez-Lapena O, Penella MT, Gasulla M. A new MPPT method for low-power solar [88] de Cesare G, Caputo D, Nascetti A. Maximum power point tracker for portable
energy harvesting. IEEE Trans Ind Electron 2010;57:3129–38. photovoltaic systems with resistive-like load. Sol Energy 2006;80:982–8.
[58] Aganah KA, Leedy AW. A constant voltage maximum power point tracking method [89] Banu IV, Beniug, x R, Istrate M. Comparative analysis of the perturb-and-observe
for solar powered systems. In: Proceedings of the 2011 IEEE 43rd Southeastern and incremental conductance MPPT methods. In: 2013 Proceedings of the 8th
symposium on system theory; 2011, pp. 125–30. international symposium on Advanced Topics in Electrical Engineering (ATEE);
[59] Leedy AW, Liping G, Aganah KA. A constant voltage MPPT method for a solar 2013. p. 1–4.
powered boost converter with DC motor load. In: Proceedings of IEEE, [90] Liu C, Wu B, Cheung R. Advanced algorithm for MPPT control of photovoltaic
Southeastcon; 2012. p. 1–6. systems. Can Sol Build Conf 2004.
[60] Yu GJ, Jung YS, Choi JY, Choy I, Song JH, Kim GS. A novel two-mode MPPT [91] Ahmed J, Salam Z. A modified P & O maximum power point tracking method with
control algorithm based on comparative study of existing algorithms. In: reduced steady state oscillation and improved tracking efficiency. IEEE Trans
Proceedings of conference record of the twenty-ninth IEEE photovoltaic specialists Sustain Energy 2016. [vol. PP, pp. 1-1].
conference; 2002, pp. 1531–4. [92] Petrone G, Spagnuolo G, Vitelli M. A multivariable perturb-and-observe maximum
[61] Shmilovitz D. "On the control of photovoltaic maximum power point tracker via power point tracking technique applied to a single-stage photovoltaic inverter.
output parameters. In: IEE Proceedings - Electric Power Applications, vol. 152; IEEE Trans Ind Electron 2011;58:76–84.
2005. p. 239–48. [93] Quamruzzaman M, Rahman KM. A modified perturb and observe maximum power
[62] David JH. Power conditioning system. U.S. patent : 3,384,806; 1968. point tracking technique for single-stage grid-connected photovoltaic inverter.
[63] Bavaro LTW. Power regulation utilizing only battery current monitoring. U.S. WSEAS Trans Power Syst 2014;9:111–8.
Patent : 4,794,272; 1988. [94] Samantara S, Roy B, Sharma R, Choudhury S, Jena B. Modeling and simulation of
[64] Li-qun L, Zhi-xin W. A rapid MPPT algorithm based on the research of solar cell’s integrated CUK converter for grid connected PV system with EPP MPPT hy-
diode factor and reverse saturation current. WSEAS Trans Syst 2008;7:568–79. bridization. In: Proceedings of 2015 IEEE Power, Communication and Information
[65] Hua C, Shen C. Study of maximum power tracking techniques and control of DC/ Technology Conference (PCITC); 2015. p. 397–402.
DC converters for photovoltaic power system. Presented at the Proceedings Power [95] Yongho K, Hyunmin J, Deokjung K. A new peak power tracker for cost-effective
Electron. Specialist Conference, Japan; 1998. photovoltaic power system. In: Proceedings of the 31st Intersociety Energy
[66] Bleijs JAM, Gow A. Fast maximum power point control of current-fed DC–DC Conversion Engineering Conference, vol. 3. IECEC 96.; 1996p. 1673–8.
converter for photovoltaic arrays. Electron Lett 2001;37:5–6. [96] Koutroulis E, Kalaitzakis K, Voulgaris NC. Development of a microcontroller-
[67] Sugimoto H, Dong H. A new scheme for maximum photovoltaic power tracking based, photovoltaic maximum power point tracking control system. IEEE Trans

23
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

Power Electron 2001;16:46–54. Electronics Specialists Conference, vol. 2. PESC 99; 1999. p. 632–7.
[97] Yeong-Chau K, Tsorng-Juu L, Jiann-Fuh C. Novel maximum-power-point-tracking [124] Chu C-C, Chen C-L. Robust maximum power point tracking method for photo-
controller for photovoltaic energy conversion system. IEEE Trans Ind Electron voltaic cells: a sliding mode control approach. Sol Energy 2009;83:1370–8.
2001;48:594–601. [125] Krein PT. Ripple correlation control, with some applications. In: Proceedings of
[98] Weidong X, Dunford WG, A modified adaptive hill climbing MPPT method for the 1999 IEEE International Symposium on Circuits and Systems, vol. 5. ISCAS '99;
photovoltaic power systems. In: Proceedings of Power Electronics Specialists 1999. p. 283–6.
Conference, Vol. 3. PESC 04. IEEE 35th Annual, 2004. p. 1957–1963. [126] Kimball JW, Krein PT. Discrete-time ripple correlation control for maximum
[99] Fangrui L, Yong K, Yu Z, Shanxu D. Comparison of P & O and hill climbing MPPT power point tracking. IEEE Trans Power Electron 2008;23:2353–62.
methods for grid-connected PV converter. In: Proceedings of the 2008 3rd IEEE [127] Casadei D, Grandi G, Rossi C. Single-phase single-stage photovoltaic generation
conference on industrial electronics and applications; 2008. p. 804–7. system based on a ripple correlation control maximum power point tracking. IEEE
[100] Peftitsis D, Adamidis G, Balouktsis A. A new MPPT method for Photovoltaic Trans Energy Convers 2006;21:562–8.
generation systems based on Hill Climbing algorithm. In: Proceedings of the 18th [128] Brunton SL, Rowley CW, Kulkarni SR, Clarkson C. Maximum power point tracking
International Conference on Electrical Machines. ICEM 2008; 2008. p. 1–5. for photovoltaic optimization using ripple-based extremum seeking control. IEEE
[101] Kjær SB. Evaluation of the "Hill Climbing" and the "Incremental Conductance" Trans Power Electron 2010;25:2531–40.
maximum power point trackers for photovoltaic power systems. IEEE Trans [129] Bodur M, Ermis M. Maximum power point tracking for low power photovoltaic
Energy Convers 2012;27:922–9. solar panels. In: Proceedings of the 7th Mediterranean Electrotechnical
[102] Thompson SP. Dynamo-electric machinery; A manual for students of electro- Conference, vol. 2; 1994. p. 758–761.
technics. London, New York: E. & F.N. Spon; 1886. [130] Noguchi T, Matsumoto H. Maximum-power-point tracking method of photovoltaic
[103] Tse KK, Ho MT, Chung HSH, Hui SY. A novel maximum power point tracker for PV power system using single transducer. In: Proceedings of the 29th Annual
panels using switching frequency modulation. IEEE Trans Power Electron Conference of the IEEE Industrial Electronics Society, Vol. 3. IECON '03;, 2003. p.
2002;17:980–9. 2350–5.
[104] Jiang J-A, Huang T-L, Hsiao Y-T, Chen2 C-H. Maximum power tracking for pho- [131] Grandi G, Rossi C, Ostojic D, Casadei D. A new multilevel conversion structure for
tovoltaic power systems. Tamkang J Sci Eng 2005;8:147–53. grid-connected PV applications. IEEE Trans Ind Electron 2009;56:4416–26.
[105] Tae-Yeop K, Ho-Gyun A, Seung Kyu P, Youn-Kyun L. A novel maximum power [132] Dhande DP, Chaudhari AP, Mahajan GK. A review of various MPPT techniques for
point tracking control for photovoltaic power system under rapidly changing solar photovoltaic system. Int J Innov Eng Res Technol (IJIERT) 2015;2.
radiation. In: Proceedings of IEEE International Symposium on Industrial [133] Matsui M, Kitano T, De-hong X, Zhong-qing Y. A new maximum photovoltaic
Electronics, vol. 2. ISIE 2001; 2001. p. 1011–4. power tracking control scheme based on power equilibrium at DC link. In:
[106] Koizumi H, Kurokawa K. A novel maximum power point tracking method for PV Proceedings of the Thirty-Fourth IAS Annual Meeting. Industry Applications
module integrated converter. In: Proceedings of the 2005 IEEE 6th power elec- Conference, vol. 2. Conference Record of the 1999 IEEE; 1999. p. 804–9.
tronics specialists conference; 2005. p. 2081–6. [134] Coelho RF, Concer FM, Martins DC. A MPPT approach based on temperature
[107] Garrigós A, Blanes JM, Carrasco JA, Ejea JB. Real time estimation of photovoltaic measurements applied in PV systems. In: Proceedings of the 9th IEEE/IAS
modules characteristics and its application to maximum power point operation. International Conference on Industry Applications (INDUSCON); 2010. p. 1–6.
Renew Energy 2007;32:1059–76. [135] Jain S, Agarwal V. A new algorithm for rapid tracking of approximate maximum
[108] Liu B, Duan S, Liu F, Xu P. Analysis and improvement of maximum power point power point in photovoltaic systems. IEEE Power Electron Lett 2004;2:16–9.
tracking algorithm based on incremental conductance method for photovoltaic [136] Kshirsagar J, Vadirajacharya K. One-cycle-controlled single-phase inverter for grid
array. In: Proceedings of the 2007 7th international conference on power elec- connected PV system. Adv Electron Electr Eng 2014;4:455–62. [2014].
tronics and drive systems; 2007. p. 637–41. [137] Yang C, Smedley KM. A cost-effective single-stage inverter with maximum power
[109] Calavia M, Perié JM, Sanz JF, Sallán J. Comparison of MPPT strategies for solar point tracking. IEEE Trans Power Electron 2004;19:1289–94.
modules. Presented at the Proceedings International Conference on Renewable [138] Femia N, Granozio D, Petrone G, Spagnuolo G, Vitelli M. Optimized one-cycle
Energies Power Quality, Granada, Spain; 2010. control in photovoltaic grid connected applications. IEEE Trans Aerosp Electron
[110] Prasad SY, Chhetri BB, Adhikary B, Bista D. Microcontroller based intelligent DC/ Syst 2006;42:954–72.
DC converter to track maximum power point for solar photovoltaic module. [139] Yu WL, Lee TP, Wu GH, Chen QS, Chiu HJ, Lo YK., et al., A DSP-based single-stage
In: Proceedings of IEEE Conference on Innovative Technologies for an Efficient maximum power point tracking PV inverter. In: Proceedings of twenty-fifth annual
and Reliable Electricity Supply (CITRES); 2010. p. 94–101. IEEE Applied Power Electronics Conference and Exposition (APEC); 2010. p.
[111] Safari A, Mekhilef S. "Incremental conductance MPPT method for PV systems. In: 948–52.
Proceedings of the 24th Canadian Conference on Electrical and Computer [140] Solodovnik EV, Shengyi L, Dougal RA. Power controller design for maximum
Engineering (CCECE); 2011. p. 000345–000347. power tracking in solar installations. IEEE Trans Power Electron
[112] Suwannatrai P, Liutanakul P, Wipasuramonton P. Maximum power point tracking 2004;19:1295–304.
by incremental conductance method for photovoltaic systems with phase shifted [141] Li X, Li Y, Seem JE, Lei P, Maximum power point tracking for photovoltaic systems
full-bridge dc-dc converter. In: Proceedings of the 8th international conference on using adaptive extremum seeking control. In: Proceedings of the 50th IEEE con-
Electrical Engineering/Electronics, Computer, Telecommunications and ference on decision and control and european control conference; 2011. p.
Information Technology (ECTI-CON), 2011. p. 637–40. 1503–8.
[113] Mirbagher SZ, Mekhilef S, Mirhassani SM. Mediterranean Green Energy Forum [142] Jamil M, Saeed H, Qaisar S, Felemban EA. Maximum power point tracking of a
2013: proceedings of an international conference MGEF-13MPPT with Inc., cond solar system using state space averaging for wireless sensor network. In:
method using conventional interleaved boost converter. Energy Procedia Proceedings of IEEE International Conference on Smart Instrumentation,
2013;42:24–32. Measurement and Applications (ICSIMA); 2013, pp. 1–6.
[114] Putri RI, Wibowo S, Rifa’i M. 2nd International Conference on Sustainable Energy [143] Khaehintung N, Pramotung K, Tuvirat B, Sirisuk P. "RISC-microcontroller built-in
Engineering and Application (ICSEEA) 2014 sustainable energy for green mobility fuzzy logic controller of maximum power point tracking for solar-powered light-
maximum power point tracking for photovoltaic using incremental conductance flasher applications. In: Proceedings of the 30th annual conference of IEEE
method. Energy Procedia 2015;68:22–30. Industrial Electronics Society, Vol. 3. IECON 2004; 2004. p. 2673–8.
[115] Rezk H, Eltamaly AM. A comprehensive comparison of different MPPT techniques [144] Karlis AD, Kottas TL, Boutalis YS. A novel maximum power point tracking method
for photovoltaic systems. Sol Energy 2015;112:1–11. for PV systems using fuzzy cognitive networks (FCN). Electr Power Syst Res
[116] Tey KS, Mekhilef S. Modified incremental conductance algorithm for photovoltaic 2007;77:315–27.
system under partial shading conditions and load variation. IEEE Trans Ind [145] Altas IH, Sharaf AM. A novel maximum power fuzzy logic controller for photo-
Electron 2014;61:5384–92. voltaic solar energy systems. Renew Energy 2008;33:388–99.
[117] Costogue EN, Lindena S. Comparison of candidate solar array maximum power [146] Syafaruddin E, Karatepe, Hiyama T. Polar coordinated fuzzy controller based real-
utilization approaches. Presented at the Intersociety Energy Conversion Eng. Conf; time maximum-power point control of photovoltaic system. Renew Energy
1976. 2009;34:2597–606.
[118] Kolluru VR, Mahapatra K, Subudhi B, Ramesh T. "Real time implementation and [147] Chiu CS, Fuzzy T-S. Maximum power point tracking control of solar power gen-
comparison of PI and modified Inc., Cond control algorithms for solar applications. eration systems. IEEE Trans Energy Convers 2010;25:1123–32.
In: Proceedings of the IEEE 6th India International Conference on Power [148] Algazar MM, Al-monier H, El-halim HA, Salem MEEK. Maximum power point
Electronics (IICPE); 2014. p. 1–6. tracking using fuzzy logic control. Int J Electr Power Energy Syst 2012;39:21–8.
[119] Liu F, Duan S, Liu F, Liu B, Kang Y. A variable step size INC MPPT method for PV [149] Rajesh R, Mabel MC. Efficiency analysis of a multi-fuzzy logic controller for the
systems. IEEE Trans Ind Electron 2008;55:2622–8. determination of operating points in a PV system. Sol Energy 2014;99:77–87.
[120] Mei Q, Shan M, Liu L, Guerrero JM. A novel improved variable step-size incre- [150] Kottas TL, Boutalis YS, Karlis AD. New maximum power point tracker for PV ar-
mental-resistance MPPT method for PV systems. IEEE Trans Ind Electron rays using fuzzy controller in close cooperation with fuzzy cognitive networks.
2011;58:2427–34. IEEE Trans Energy Convers 2006;21:793–803.
[121] Hsieh GC, Hsieh HI, Tsai CY, Wang CH. Photovoltaic power-increment-aided in- [151] Shiau J-K, Wei Y-C, Chen B-C. A study on the fuzzy-logic-based solar power MPPT
cremental-conductance MPPT With two-phased tracking. IEEE Trans Power algorithms using different fuzzy input variables. Algorithms 2015;8:100.
Electron 2013;28:2895–911. [152] Hiyama T, Kitabayashi K. Neural network based estimation of maximum power
[122] Hohm DP, Ropp ME, "Comparative study of maximum power point tracking al- generation from PV module using environmental information. IEEE Trans Energy
gorithms using an experimental, programmable, maximum power point tracking Convers 1997;12:241–7.
test bed. In: Proceedings of photovoltaic specialists conference. Conference record [153] Huang K, Li W, Huang X. MPPT of solar energy generating system with fuzzy
of the twenty-eighth IEEE, 2000. p. 1699–702. control and artificial neural network. In: Proceedings of International Conference
[123] Brambilla A, Gambarara M, Garutti A, Ronchi F. "New approach to photovoltaic on Information Technology, Computer Engineering and Management Sciences
arrays maximum power point tracking. In: Proceedings of 30th Annual IEEE Power (ICM); 2011. p. 230–3.

24
M.A. Danandeh, S.M. Mousavi G. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx–xxx

[154] Seyedmahmoudian M, Horan B, Soon TK, Rahmani R, Than Oo AM, Mekhilef S, [173] Miyatake M, Veerachary M, Toriumi F, Fujii N, Ko H. Maximum power point
et al. State of the art artificial intelligence-based MPPT techniques for mitigating tracking of multiple photovoltaic arrays: a PSO approach. IEEE Trans Aerosp
partial shading effects on PV systems – a review. Renew Sustain Energy Rev Electron Syst 2011;47:367–80.
2016;64:435–55. [174] Ishaque K, Salam Z, Amjad M, Mekhilef S. An improved particle swarm optimi-
[155] Gheibi A, Mohammadi SMA, maghfoori M. The Proceedings of International zation (PSO)– based MPPT for PV with reduced steady-state oscillation. IEEE Trans
Conference on Smart Grid and Clean Energy Technologies (ICSGCE 2011) max- Power Electron 2012;27:3627–38.
imum power point tracking of photovoltaic generation based on the type 2 fuzzy [175] Soon JJ, Low KS. Photovoltaic model identification using particle swarm optimi-
logic control method. Energy Procedia 2011;12:538–46. zation with inverse barrier constraint. IEEE Trans Power Electron
[156] Kassem AM. MPPT control design and performance improvements of a PV gen- 2012;27:3975–83.
erator powered DC motor-pump system based on artificial neural networks. Int J [176] Mirhassani SM, Razzazan M, Ramezani A. An improved PSO based MPPT approach
Electr Power Energy Syst 2012;43:90–8. to cope with partially shaded condition. In: Proceedings of the 22nd Iranian
[157] Kulaksiz AA, Aydoğdu Ö. ANN-based maximum power point tracking of photo- Conference on Electrical Engineering (ICEE); 2014. p. 550–5.
voltaic system using fuzzy controller. In: Proceedings of International Symposium [177] Chao K-H, Lin Y-S, Lai U-D. Improved particle swarm optimization for maximum
on Innovations in Intelligent Systems and Applications (INISTA); 2012. p. 1–5. power point tracking in photovoltaic module arrays. Appl Energy
[158] Farhat S, Alaoui R, Kahaji A, Bouhouch L. Estimating the photovoltaic MPPT by 2015;158:609–18.
artificial neural network. In: Proceedings of International Renewable and [178] Renaudineau H, Donatantonio F, Fontchastagner J, Petrone G, Spagnuolo G,
Sustainable Energy Conference (IRSEC); 2013. p. 49–53. Martin JP, et al. A PSO-based global MPPT technique for distributed PV power
[159] Gupta A, Kumar P, Pachauri RK, Chauhan YK. Performance analysis of neural generation. IEEE Trans Ind Electron 2015;62:1047–58.
network and fuzzy logic based MPPT techniques for solar PV systems. In: [179] Sudhakar Babu T, Rajasekar N, Sangeetha K. Modified Particle swarm optimiza-
Proceedings of the 6th IEEE Power India International Conference (PIICON); 2014. tion technique based maximum power point tracking for uniform and under
p. 1–6. partial shading condition. Appl Softw Comput 2015;34:613–24.
[160] Singh MD, Shine VJ, Janamala V. Application of artificial neural networks in [180] Wang Y, Bian N. Research of MPPT control method based on PSO algorithm.
optimizing MPPT control for standalone solar PV system. In: Proceedings of In: Proceedings of the 4th International Conference on Computer Science and
International Conference on Contemporary Computing and Informatics (IC3I); Network Technology (ICCSNT); 2015. p. 698–701.
2014. p. 162–6. [181] Oliveira FMd, Silva SAOd, Durand FR, Sampaio LP, Bacon VD, Campanhol LBG.
[161] Anzalchi A, Sarwat A. "Artificial neural network based duty cycle estimation for Grid-tied photovoltaic system based on PSO MPPT technique with active power
maximum power point tracking in photovoltaic systems. In: Proceedings of line conditioning. IET Power Electron 2016;9:1180–91.
SoutheastCon 2015, 2015. p. 1–5. [182] Saad NH, El-Sattar AA, Mansour AE-AM. Improved particle swarm optimization
[162] Elobaid LM, Abdelsalam AK, Zakzouk EE. Artificial neural network-based photo- for photovoltaic system connected to the grid with low voltage ride through
voltaic maximum power point tracking techniques: a survey. IET Renew Power capability. Renew Energy 2016;85:181–94.
Gener 2015;9:1043–63. [183] Rania H, Babak C, Olivier de W, Gerhard V. "A comparison of particle swarm
[163] Kurniawan A, Haryanto E, Masroeri AA. A neural network based maximum power optimization and the genetic algorithm. In: Proceedings of the 46th AIAA/ASME/
point tracker with KY converter for photovoltaic system on a moving vehicle. In: ASCE/AHS/ASC structures, structural dynamics and materials conference, ed:
Proceedings of International Conference on Advanced Mechatronics, Intelligent American Institute of Aeronautics and Astronautics; 2005.
Manufacture, and Industrial Automation (ICAMIMIA), 2015, pp. 117–120. [184] Ramaprabha R, Mathur BL. Genetic algorithm based maximum power point
[164] Messalti S, Harrag AG, Loukriz AE. A new neural networks MPPT controller for PV tracking for partially shaded solar photovoltaic array. Int J Res Rev Inf Sci (IJRRIS)
systems. In: Proceedings of the 6th International Renewable Energy Congress 2012;2.
(IREC), 2015, pp. 1–6. [185] Daraban S, Petreus D, Morel C. A novel MPPT (maximum power point tracking)
[165] Ramana VV, Jena D. "Maximum power point tracking of PV array under non- algorithm based on a modified genetic algorithm specialized on tracking the global
uniform irradiance using artificial neural network. In: Proceedings of IEEE inter- maximum power point in photovoltaic systems affected by partial shading. Energy
national conference on Signal Processing, Informatics, Communication and Energy 2014;74:374–88.
Systems (SPICES); 2015. p. 1–5. [186] Amrouche B, Belhamel M, Guessoum A. Artifical intelligence based P & O MPPT
[166] Rizzo SA, Scelba G. ANN based MPPT method for rapidly variable shading con- method for photovoltaic systems. Presented at the Proceedings Rev. Energies
ditions. Appl Energy 2015;145:124–32. Renouvelables, Tlemcen, Algeria; 2007.
[167] Chine W, Mellit A, Lughi V, Malek A, Sulligoi G, Pavan A Massi. A novel fault [187] Ma J, Man KL, Ting TO, Zhang N, Lei CU, Wong N. A hybrid MPPT method for
diagnosis technique for photovoltaic systems based on artificial neural networks. Photovoltaic systems via estimation and revision method. In: Proceedings of 2013
Renew Energy 2016;90:501–12. IEEE International Symposium on Circuits and Systems (ISCAS2013); 2013. p.
[168] Makhloufi MT, Abdessemed Y, Khireddine MS. A neural network MPP tracker 241–4.
using a Buck-Boost DC/DC converter for photovoltaic systems. In: Proceedings of [188] Ramprava R, Mathur BL. Intelligent controller based maximum power point
the 5th International Conference on Systems and Control (ICSC); 2016. p. 72–7. tracking for solar PV system. Int J Comput Appl - IJCA 2011;12:37–41.
[169] Yang XS. Nature-inspired metaheuristic algorithms. Beckington: Luniver Press; [189] Larbes C, Aït Cheikh SM, Obeidi T, Zerguerras A. Genetic algorithms optimized
2008. fuzzy logic control for the maximum power point tracking in photovoltaic system.
[170] Yang X-S. Firefly algorithms for multimodal optimization. In: Watanabe O, Renew Energy 2009;34:2093–100.
Zeugmann T, editors. Proceedings of stochastic algorithms: foundations and ap- [190] Harrag A, Messalti S. Variable step size modified P & O MPPT algorithm using GA-
plications: 5th international symposium, SAGA 2009, Sapporo, Japan, October 26- based hybrid offline/online PID controller. Renew Sustain Energy Rev
28, 2009. Berlin, Heidelberg: Springer Berlin Heidelberg; 2009. p. 169–78. 2015;49:1247–60.
[171] Sundareswaran K, Peddapati S, Palani S. MPPT of PV systems under partial shaded [191] Peng BR, Chen JH, Liu YH, Chiu YH. Comparison between three different types of
conditions through a colony of flashing fireflies. IEEE Trans Energy Convers variable step-size P & O MPPT technique. In: Proceedings of international con-
2014;29:463–72. ference on Computer Information Systems and Industrial Applications (CISIA
[172] Fu Q, Tong N. A complex-method-based PSO algorithm for the maximum power 2015); 2015.
point tracking in photovoltaic system. In: Proceedings of the second international [192] Badis A, Mansouri MN, Sakly A. PSO and GA-based maximum power point
conference on Information Technology and Computer Science (ITCS); 2010. p. tracking for partially shaded photovoltaic systems. In: Proceedings of 2016 7th
134–7. International Renewable Energy Congress (IREC); 2016. p. 1–6.

25