MPPT for Photovoltaic System Using Multi-objective Improved Particle Swarm Optimization Algorithm

TELKOMNIKA Indonesian Journal of Electrical Engineering
Vol.12, No.1, January 2014, pp. 261~ 268
DOI: http://dx.doi.org/10.11591/telkomnika.v12i1.3965  261
Received June 21, 2013; Revised August 13, 2013; Accepted September 16, 2013
MPPT for Photovoltaic System Using Multi-objective
Improved Particle Swarm Optimization Algorithm
Yuhong Zhao*1
, Xuecheng Zhao2
, Yunhui Zhang3
1
School of Electric Engineering, University of South China
Hengyang, 421001, Hunan, China; PH./Fax: 0734-8282044
2
Mechanical and Electrical Engineering Department, Shaoyang Vocational and Technical College
Shaoyang, 422000, Hunan, China;
3
Mechanical and Electrical Engineering Department, Shenzhen Polytechnic
Shenzhen, 518057, Guangdong, China;
*Corresponding author, e-mail: gsxl666@163.com
1
, ljzy666@163.com
2
, zhangyunhui1106@163.com
3
Abstract
Making full use of abundant renewable solar energy through the development of photovoltaic (PV)
technology is an effective means to solve the problems such as difficulty in electricity supply and energy
shortages in remote rural areas. In order to improve the electricity generating efficiency of PV cells, it is
necessary to track the maximum power point of PV array, which is difficult to make under partially shaded
conditions due to the odds of the appearance of two or more local maximum power points., In this paper, a
control algorithm of maximum power point tracking (MPPT) based on improved particle swarm optimization
(IPSO) algorithm is presented for PV systems. Firstly, the current in maximum power point is searched
with the IPSO algorithm, and then the real maximum power point is tracked through controlling the output
current of PV array. The MPPT method based on IPSO algorithm is established and simulated with Matlab
/ Simulink, and meanwhile, the comparison between IPSO MPPT algorithm and traditional MPPT algorithm
is also performed in this paper. It is proved through simulation and experimental results that the IPSO
algorithm has good performances and very fast response even to partial shaded PV modules, , which
ensures the stability of PV system.
Keywords: photovoltaic system, maximum power point tracking, improved particle swarm optimization
Copyright © 2014 Institute of Advanced Engineering and Science. All rights reserved.
1. Introduction
Due to continuous growth of global energy demand and increasing concern about
environmental issues, interests on using and developing renewable energy sources are
growing.The solar energy is known to be one of the preferred renewable green energies, which
is much cleaner and free from harmful production to environment compared with the
conventional counterparts [1]. In this paper, improved MPPT algorithm for solar PV technology
is researched to solve the difficulty in electricity consumption in remote rural areas.
The solar cell is the core component of the solar photovoltaic system, and the power
produced by a solar cell is dependent on the solar irradiation and the temperature of the solar
cell. Its voltage shows the complex non-linear, which leads to the instability of the output power
and the lower power generation efficiency. Therefore, how to achieve effective MPPT control
and further improve the conversion efficiency of the PV array and make full use of its energy
conversion has been an important topic of PV power generation system. At present, the
common control methods both in domestic and foreign include perturbation and observation
method, a constant voltage method, incremental conductance method, and so on [2]. Due to the
partial obstructions such as surrounding buildings, trees, the part of PV array is partially covered
with shadow, in which case, the PV array output characteristic shows multi-local maximum
operating point and thus the conventional MPPT control algorithm will fail to track the real
maximum power point. To solve this problem, an improved MPPT technique has been put
forward by means of particle swarm optimization algorithm for Photovoltaic Grid Connected
Generation system. The algorithm with a good performance in multi-objective optimization and
multimodal function optimization is suitable for the shaded photovoltaic system [3], [4].

 ISSN: 2302-4046
TELKOMNIKA Vol. 12, No. 1, January 2014: 261 – 268
262
2. Improved Particle Swarm Optimization Algorithm Design
2.1. The Principle of the Particle Swarm Optimization Algorithm
Particle swarm optimization (PSO) is an evolution computing technology based on
swarm intelligence, which was proposed by Eberhart and Kennedy in 1995, and it is based on
the simulation of the flock. The principle is described as follows: a flock of birds search food
randomly in a region where there is only a piece of food. The most simple and effective strategy
to find food is to search the area around the nearest bird from food, which constitutes one of the
basic concepts of the PSO. Assuming that a community is composed of n particles in a D-
dimensional search space, wherein xi is the position of i-th particle in D-dimension [5], [6].
 1 2, , , 1,2,i i i idX X X X i n   (1)
Fitness value can be calculated by substituting Xi into an objective function, and the
pros and cons of Xi can be obtained according to the size of the fitness value. "Flying" velocity
of the particle i is a D-dimensional vector, denoted as vi = (vi1, vi2... vid). Pi = (pi1, pi2, ⋯, pid,) is the
best position of particle i searched, corresponding to the optimal solution particle i found by
itself, which is the location of the best fitness value. Optimal position for the entire particle
swarm search after the h-th iteration is Pigd = (Pi1, Pi2... Pid). By the above definition, the
standard formula of PSO can be expressed as:
( 1) ( ) ( ) ( ) ( ) ( ) ( ) ( )
1 1 2 2( ) ( )k k k k k k k k
id id id id gd idV v c r p x c r p x
     (2)
( 1) ( ) ( 1)k k k
id id idx x V 
  (3)
where i is the number of particles in the swarm; d represents the dth-dimension of
particles; ω is the inertia weight coefficient; r1, r2 are the random values between [0, 1]; c1 called
cognitive factor represents belief degree on experience, which can be used to adjust the step
size of particles to fly towards the direction of its local best position; c2 known as the coefficient
of social learning represents the belief degree on individuals around, which can be used to
adjust the step size of particles to fly towards the direction of its global best position. The
algorithm iteration termination condition is generally chosen as the maximum number of
iterations or fitness value which satisfies the predetermined threshold value of the minimum
fitness after searching the optimal location.
PSO algorithm is easy to operate and simple for use with a fast convergence. However,
the algorithm also has the following problems:
1) The particles are "flying" toward the direction of the optimal solution. However, if the inertia
factor is large, it is difficult to obtain the optimal solution, and the search accuracy will
reduce;
2) All the particles are "flying" toward the direction of the optimal solution, but the closer the
optimal particle comes, the less its searching speed becomes. Particle swarm deprives the
diversity of solutions between the particles, and thus the algorithm may converge to a local
maximum without difficulty which is not always the same as the global maximum and fails to
track the actual global maximum.
2.2. Improved Particle Swarm Optimization (IPSO) Algorithm
Some modifications are proposed to improve the search algorithm: In the direction of
the same velocity, new particles are classified according to different amplitudes. Global
optimization is performed on particles with larger velocity amplitude, while local optimization is
performed on the other particles. Individuals and global best position "extreme value" are
selected to update the particle velocity. Large velocity amplitude which meets the particle global
search requirements can prevent the algorithm from falling into local optimal and premature
phenomenon, and small velocity amplitude which meets local search requirements can avoid
the algorithm from overflying the optimal solution space, and then the optimum solution can be
obtained faster. [4] Formula is shown as follows:

TELKOMNIKA ISSN: 2302-4046 
MPPT for Photovoltaic System Using Multi-objective Improved Particle Swarm... (Yuhong Zhao)
263
( 1) ( ) ( ) ( ) ( ) ( ) ( ) ( )
1 1 2 2
( 1) ( ) ( 1)
( ) (0)
( ) (0) ( )
( ) ( )
( ) , 1,2 ,
, 1,2 ,
k k k k k k k k
id id id id gd id
k k k
id id id
n
id id
n n
id id id
v v c p x c p x
x x V
v a n v n j
x x v n j
  
 
     

 

 

  


(4)
where i is the number of particles; d represents the dth-dimension of particles; ω is the
inertia weight coefficient; r1, r2 are the random values between [0, 1]; (0)
idv is called the dth-
dimensional reference velocity component of particle i ; ( )n
idv ,n =1,2,…, j is called the dth-
dimension search velocity component of particle i, j is called speed interval number; (0)
idx is called
the dth-dimension reference position component of particle i, ( )n
idx ,n =1,2,…, j is called the dth-
dimension search position component of particle i; idp is the best position of particle i that has
ever found; the best location found in the group is called the global best position, recorded as
gdp . a(n), n = 1, 2... j , called speed coefficient, is used to determine the relationship between
the search speed and the reference speed. The method to determine the relationship between
the search speed and the reference speed is to set a maximum speed maxv and a minimum
speed minv . If (0)
maxidv v , then a(n) search speed is small; if (0)
minidv v , then a(n) search speed
becomes large; if (0)
min maxidv v v  , and when search speed is appropriate, a(n) to the search
speed (0)
idv of both sides becomes large or smaller?. Only in this way, enough solution space can
be searched. Therefore, the formula is:
(0)
min
(0)
max
(0)
min max
,
( ) / ,
1 /
id
id
id
n v v
a n n j v v
n j v v v
 

 

  
(5)
The study shows that the inertia factor ω has a significant impact on optimize
performance of the algorithm. Large ω will improve the convergence rate of the algorithm. While
ω is small, it will improve the convergence accuracy of the algorithm. According to formula (6),
this paper proposed an inertia weight adaptive adjustment strategy.
1
1
ex p ( / ) 1
*
i
i
i n
i
i
a F
n b
D F




  

(6)
Where: D is the particle dimension; n is the number of particles; ωi is the i-th particle
inertia factor; Fi is the i-th particle fitness; a、b is the two adjustable parameters. We could
adjust the inertia weight based on experience and combined simulation, the experiments show
that the inertia weight adaptive adjustment strategy has a better effect for multi-peak function
optimization.
2.3. MTTP has been Design using Improved Particle Swarm Optimization Algorithm
In this paper, two blocks photovoltaic panels connected in series are used for example
to analyze the process of algorithm design. The algorithm design principle of more blocks PV
modules connected in series model is the same, that is, just change the number of relevant
variables [7]. The program flowchart of the IPSO algorithm introduced in this paper is shown in
Figure 1.

 ISSN: 2302-4046
264
The specific procedure is as follows:
① Model transformation
Multi-peak MPPT control model of photovoltaic system and PSO model are different,
and the transformation relation is needed to build. The global optimal location is the maximum
power output from two pieces of tandem photovoltaic module in the control object, and the
speed of the particles is the output current of the tandem photovoltaic array in the control object.
② Initialization of the population parameter
Population number and the number of these individuals in the group are initialized;
evolution algebra is set; the parameters of the learning factor, the weighting coefficients and
maximum speed are initialized; each particle's initial position and velocity of each particle are
given in a random way
③ Calculation of the fitness value of each particle
After various parameters have been initialized, the adaptation value of the objective
function for each particle is calculated. The objective function is the total power output of the
array, and the fitness function expression is:
( , 1)* ( , 2)*power V I Sun V I Sun I  (7)
Where I is the output current of the PV array, V (I, Sun1) represents the output voltage
of the photovoltaic cell module when the light intensity is Sun1 and the output current is I.
Figure 1. Flowchart of the IPSO algorithm

265
④ Find individual optimal location
Every individual in evolution has an optimal position in the individual history, namely
individual extremum. This individual fitness is compared with the fitness value of the particles
(the maximum power of the system). If current value is better, then the position with current
position as well as the fitness value is updated simultaneously.
⑤ Find the global optimal position
Each individual has a history optimal location. For the whole group, there is also a
global optimal position. The best particle of group with the best fitness value is determined. The
optimal position is the position of the optimal individual in the population.
⑥ Update the particle velocity and position
The speed of particle swarm model is equivalent to actual model of the current
increment. Position is equivalent to the current of the actual model. The position and velocity of
each particle are updated according to formula (1) and (2).
⑦ Determine the termination condition
End conditions are maximum evolution generation. The search is terminated when the
iteration becomes the maximum evolution generation. Otherwise, the evolution algebra pluses 1
automatically, and returns to step 2. The optimal solution Imax is the current value at the
maximum power point.
3. MPPT Simulation using Improved Particale Swarm Algorithm Simulation for
Photovoltaic Systems
3.1. Simulation Control Model
By means of a simulation model of photovoltaic grid-connected systems developed in
MATLAB/SUMULINK software, the actual control effect of the improved algorithm proposed in
this paper has been verified, and the system structure is shown in Figure 2. The MPPT is
controlled by DC (direct current) / DC and this DC can then be converted into AC (alternating
current) by DC/AC inverter. The inverter is controlled by DC / AC to obtain stable DC voltage.
The particle swarm algorithm is responsible for finding the output current of the PV array at the
maximum power point, and hysteresis comparison has been made between the current and the
actual current to generate the PWM signal to control the output current of the PV array [8], [9].
Figure 2. The configuration diagram of photovoltaic grid-connected systems based on improved
particle swarm optimization MPPT algorithm
3.2. Simulation Parameter Initialization
Particle Swarm initializations are as follows [10]:
pop_size=20;
part_size=2;
max_gen=80;
w_max=0.9;
w_min=0.3;
v_max=2;
c1=2;
c2=2;

 ISSN: 2302-4046
266
Light intensity initializations are as follows:
S1=S2=1KW/m2
S3=S4=0.8
KW/m2
Hysteresis loop width is set as follows:
0.01（），－（）on 0.01 off
DC bus voltage is given as follows:
400V
3.3. Simulation Results
The simulation results are shown in Figure 3, 4, 5. Figure 3 shows the variation curves
of output voltage and current of the PV array with time. The system is stable at about 0.02s.
Output voltage of the PV array is about 117.1V in steady state, and the fluctuation range of
voltage is ±0.5V. Output current is stable at 7.2A, which is almost equal to the theoretical
calculation value. The fluctuation range of current is ±0.04A. Figure 4 reflects the input voltage
curve of the inverter under voltage feedback control, which is stable at 400V when the system is
stable at 0.28s, and the fluctuation range of voltage value is ± 0.6V in steady state. Figure 5
shows the voltage and current waveforms in steady state. It can be seen that grid current and
voltage vary with frequency and in phase (reference direction reverse) in steady state, and the
electricity can be utilized with safety.
Figure 3. The variation curves of the PV array output voltage and current with time
Figure 4. Inverter DC voltage dynamic curve

267
Figure 5. The variation curve of grid voltage and grid current with time
The IPSO algorithm is superior to traditional algorithm due to its ability to search the
global maximum, which can be applied in MPPT for the shaded PV system. In order to verify the
superiority of the IPSO algorithm, a MPPT simulation of a PV power supply system with the
proposed IPSO algorithm is implemented and compared with the results obtained from
traditional incremental conductance algorithm.
Figure 6. Power-time curve obtained by incremental conductance algorithm
Figure 7. Power - time curve obtained by improved particle swarm algorithm
Figure 6 and Figure 7 show the variation curves of the system power with time by using
the improved particle swarm MPPT algorithm and conventional incremental conductance MPPT
algorithm, respectively. It can be seen that the maximum power of 682 w obtained from the
improved particle swarm MPPT algorithm is higher than that of 607 w obtained by the

 ISSN: 2302-4046
268
incremental conductance method. The improved particle swarm algorithm can track the
maximum power point in 0.03 s, but the time incremental conductance method spends is 0.13 s.
In steady state, the fluctuation range of the power obtained from the incremental conductance
algorithm is ±8W, but that from the IPSO algorithm is ±1W. It can be seen that the method can
not only improve the tracking speed of MPPT, but also ensure the tracking accuracy.
4. Conclusion
In remote areas far from the power grid where are abound with solar energy resources,
electricity supply is severe,. Therefore, making full use of abundant solar energy resources
realized by the development of photovoltaic grid technology is an effective means to solve the
difficulties in rural electricity supply, energy shortages and other issues. This paper proposes a
MPPT algorithm using improved particle swarm optimization algorithm for PV system, which can
solve the problem that the traditional MPPT algorithm is liable to fall into the local optimal
solution under partially shaded conditions. This paper has established a simulation model of the
proposed algorithm. Simulated results have demonstrated the feasibility of the IPSO algorithm
and the IPSO is suitable for MPPT in grid-connected PV systems. By comparing the traditional
method with the IPSO method, the experiment has verified the advantages of the proposed
IPSO algorithm: it has a faster dynamic response and better steady-state performance than the
traditional algorithm, thus improving the efficiency of the photovoltaic power generation system.
Acknowledgements
This work was supported by the Scientific Research Fund of Hunan Provincial Science
and Technology Programme (Project Number [2014FJ3157]) and Hengyang City Science and
Technology Programme (Project Number [2013KJ31]).
References
[1] Aryuanto Soetedjo. Modeling of Maximum Power Point Tracking Controller for Solar Power System.
TELKOMNIKA Indonesian Journal of Electrical Engineering. 2012; 10(3): 419-430.
[2] Esram T, Chapman PL. Comparison of photovoltaic array maximum power point tracking techniques.
IEEE Transactions on Energy Conversion l. 2007; 22(2): 439-449.
[3] Ramaprabha R, Mathur BL. A comprehensive review and analysis of solar photovoltaic array
configurations under partial shaded conditions. Int J Photo Energy. 2012; 28(2): 1-16.
[4] Ramaprabha R, Mathur BL. Soft computing optimization techniques for solar photovoltaic arrays.
ARPN J Eng Appl Sci. 2011; 6(10):70-5.
[5] JL Fernández Martínez, E García Gonzalo. The PSO family: deduction, stochastic analysis and
comparison. Swarm Intelligence. 2009; 3(4): 245-273.
[6] Omkar SN, Mudigere D, Narayananaikg, et al. Vector evaluated particle swarm optimization formulti-
objective design optimization of composite structures. Computers and Structures. 2008; 86(2): 1-14.
[7] Hua C, Lin J. A modified tracking algorithm for maximum power tracking of solar array. Energy
Conversion and Management. 2004; 45(6): 911-925.
[8] Zhou Tianpei. MPPT Method of Wind Power Based on Improved Particle Swarm Optimization.
TELKOMNIKA Indonesian Journal of Electrical Engineering. 2013; 11(6): 3206-3212.
[9] Shimizu T, Wada K, Nakamura N. Fly back-type single-phase utility interactive inverter with power
pulsation decoupling on the DC input for an AC photovoltaic module system. IEEE Trans. on Power
Electronics. 2006; 21(5): 1264-1272.
[10] Celik AN, Acikgoz N. Modeling and experimental verification of the operating current of mono-
crystalline photovoltaic modules using four and five-parameter models. Applied Energy. 2007; 84(5):
11-25.

MPPT for Photovoltaic System Using Multi-objective Improved Particle Swarm Optimization Algorithm

Recommended

More Related Content

What's hot (19)

Similar to MPPT for Photovoltaic System Using Multi-objective Improved Particle Swarm Optimization Algorithm (20)

More from Nooria Sukmaningtyas (20)

Recently uploaded (20)

MPPT for Photovoltaic System Using Multi-objective Improved Particle Swarm Optimization Algorithm