International Conference on Innovations in Power and Advanced Computing Technologies [i-PACT2017]
Maximum Power Point Tracking for PV Panels
using Ant Colony Optimization
Sarat Kumar Sahoo, Balamurugan M, Sai Anurag, Rohan Kumar, Vishnu Priya
School of Electrical Engineering, VIT University, Vellore, India
[email protected],
[email protected],
[email protected],
[email protected],
[email protected]Abstract— Output power of solar panel varies with respect to Io -Reverse saturation current
irradiance and temperature. So maximum power should be V -Cell voltage
extracted from solar panels using Maximum Power Point Rs -Series resistance
Tracking. As irradiance changes continuously during the day, Rp -Parallel resistance
many algorithms fail to give maximum power as they get stuck at VT-Thermal voltage = KT/q
local maximum. In order to overcome this problem, we propose
K-Boltzman constant
Ant Colony Optimization which obtains global maximum and
improves the efficiency of solar panels. The proposed work has T-Temperature in Kelvin
been implemented using Matlab/Simulink. Also, the hardware q-Charge of an electron
results obtained indicate the superiority of Ant Colony
Optimization over the conventional Maximum Power Point Eq.1 shows that the output current depends upon
Tracking methods. irradiance and temperature. Output power (P) of PV cell varies
non-linearly with respect to voltage (V) and it has only one
maximum power point [10]. Practically, irradiance will not be
Keywords— Maximum Power Point Tracking (MPPT); Ant Colony constant and there are problems of partial shading too [2]. This
Optimization (ACO); Photovoltaic (PV); Boost Converter
results in multiple peaks which makes it difficult to track the
maximum power point [5]. Also, the efficiency of solar panels
I. INTRODUCTION is very low which makes MPPT more essential.
Due to over exploitation and depletion of fossil fuels over There are many other algorithms [3,4], [7,8] like Perturb and
the past few years, demand for the renewable energy resources Observe (P&O) [9], Incremental Conductance which are more
has been increasing tremendously. Among the available popularly used these days. But these algorithms give best
renewable energy resources, solar energy is considered to be results in uniform irradiance and get stuck at local maxima
the best because of the various advantages it has (easily when irradiance is not constant. So there is a need for efficient
available; free of cost; non-polluting) [6]. algorithm which can overcome all these constraints.
This paper proposes Ant Colony Optimization (ACO) to
A. PV Model solve the problem and the results obtained were satisfactory.
I. ANT COLONY OPTIMIZATION
Ant Colony Optimization is generally used to solve non-
linear problems. ACO mimics the behavior of ants to find the
optimized path. Initially ants move randomly in different
directions in search of food. They deposit pheromone on their
way for other ants to follow. The deposited pheromone can
also evaporate as time passes, so the probability of finding
Fig. 1. Equivalent circuit model of a PV cell
pheromone is more for the shortest path. This process is
repeated for a number of iterations so that optimized path can
be found.
V + I .R s V + I .R s Consider a problem in which A parameters are to be
I = I ph − I o .[exp( ) − 1] − [ ] (1) optimized. Initially Y random solutions where YA are
VT Rp generated. Also new solutions are formed by sampling
Gaussian Kernel using the following Eq.2
I -Cell current
Iph -Insolation current
978-1-5090-5682-8 /17/$31.00 ©2017 IEEE
1
International Conference on Innovations in Power and Advanced Computing Technologies [i-PACT2017]
Step 6: The corresponding current and voltage for obtained
y y
1 ( x − μli ) 2
Gi ( x ) = ¦ ωl gli ( x ) = ¦ ωl exp( − 2
) (2) maximum power is noted. Duty cycle is calculated
l =1 l =1 σ li 2 ∏ 2σ li and is given to the boost converter.
Gi (x) - Gaussian Kernel for ith dimension
Step 7: Steps 2-6 will be repeated if there is a large change in
g li (x ) - lth sub Gaussian function for the ith dimension Irradiance.
σ li and μ li are the ith dimensional standard deviation and
mean values for the lth solution.
Using Y initial solutions, mean, standard deviation and
weights are calculated as follows:
Mean:
μ i = {μ1i ,......μ li ,...μ yi } = {s1i ,.....s li ,.......s iy } (3)
Standard Deviation:
y s ij − s li
σ = ξ¦
i
l
(4)
j =1 Y −1
Weight:
1 (l − 1) 2 (5)
ωl = exp(− )
QK 2 ∏ 2Q 2 K 2
ξ - Convergence rate
The constants are taken from [1]
The probability of selecting lth Gaussian function can be
calculated by
ωl (6)
pl = y
¦ωr =1
r
This sampling process will be repeated according to the
number of parameters to be optimized. Now Z new solutions
that are generated are added to the Y initial solutions. These
Z+Y solutions are ranked and Y best solutions are restored
again. The entire process is repeated for the required number
of iterations.[1] Fig. 2. Flow chart of ACO for MPPT
A. Steps involved II. IMPLEMENTATION OF ACO FOR MPPT
Step 1: Initialize the parameters (Max_Ite, Z, Y, Q, ξ )
Step 2: Obtain voltage, current and calculate power for each
ant. Repeat the same till Y ants.
Step 3: Obtain Z new solutions using Gaussian function.
Step 4: Rank M+K solutions and store best K solutions.
Step 5: The above four steps are repeated till maximum
number of iterations.
2
International Conference on Innovations in Power and Advanced Computing Technologies [i-PACT2017]
Fig. 3. Control Design of MPPT using ACO. The simulation results are shown in Fig.4 and Fig.5. It
shows that Maximum Power Point can be tracked accurately
The basic block diagram of a PV system is shown in Fig.3. using ACO algorithm. The output is taken under constant
The algorithm mentioned in the paper has been tested in irradiance and temperature. In real time, this is not the case.
Matlab/Simulink and is implemented in hardware. Following hardware results show the performance of algorithm
in practical case.
A. Simulation Results B. Hardware Setup
The Simulink model consists of a DC-DC Boost converter
whose switching frequency is 20 kHz. Sampling time is 20 μs.
The values of inductor and capacitor are 25μH and 1100μF
respectively. The type of load used is resistive. The simulation
is carried out under constant irradiance (1000 W/m²) and
temperature (25ºC).
Fig. 6. Hardware setup showing MPPT control
The type of solar panel used is Polycrystalline 250W
(EMMVEE). Load that is used is resistive(3A, 67.5ohms).
Current and Voltage probes are used to handle the high
voltage and current intake and they are measured using
sensors(LEM 25-VP,LEM 55-A).
Implementation of the algorithm is done using D-Space
1103 which acts as an interface between hardware and the
Fig. 4. Output waveforms of Voltage, Current and Power of PV panel
Matlab/Simulink [11].
C. Hardware Results
Fig. 5. Input and output waveforms of Boost converter
Fig. 7. Input and output waveforms of Boost converter
3
International Conference on Innovations in Power and Advanced Computing Technologies [i-PACT2017]
It can be observed that the algorithm works well in real REFERENCES
time where irradiance and temperature are not constant. The
output power is found to be almost equal to that of simulation
results. The setup is tested for two different loads and the [1]. Lian Lian Jiang, Douglas L.Maskell, Jagdish C.Patra, “A novel ant
results are satisfactory. colonyy optimization-based maximum power point tracking for
photovoltaic systems under partially shaded conditions”, L.L. Jiang et
al. / Energy and Buildings 58 (2013) 227–236.
[2]. L. Gao, R. A. Dougal, S. Liu, and A. P. Iotova, “Parallel-connected
solar PV P system to address partial and rapidly fluctuating shadow
conditions,” IEEE Trans. Ind. Electron., vol. 56, no. 5, pp. 1548–1556,
May 2009 .
[3]. Sumathi.A, Samhita.R, Sugumar. S, Sarat Kumar Sahoo, Raja Das “A
Novel Maximum Power Point Tracking Method for Pv
Systems,”International Conference on Science, Engineering and
Technology, 2014.
[4]. J. Prakash, Sarat Kumar Sahoo “ Modeling of photovoltaic system and
design of MPPT controller using PSO algorithm," International
conference on Environment & energy ,pp.36-41, 2013.
[5]. R Sridhar, Pradeep Vishnuram D, Himabindu A, Divya, “Ant colony
optimization based maximum power point tracking (MPPT) for partially
shaded standalone PV system”, International science press, 2016.
[6]. Sarat Kumar Sahoo, Mayur Bansal “ MPPT Technique-A Review,”
International Conference on Energy and Sustainability,pp.182-187,2014.
Fig. 8. Input and output waveforms of Boost converter
[7]. Balamurugan.M, Narendiran.S, Sarat Kumar Sahoo, Ashwin Kumar
Sahoo, “Application of Particle Swarm Optimization for Maximum
Power Point Tracking in PV System”, 2016 3rd International
Conference on Electrical Energy Systems.
III. CONCLUSION
ACO algorithm for MPPT has been implemented in this [8]. Jiang, Lian Lian, D.R. Nayanasiri, Douglas L. Maskell, and D.M.
Vilathgamuwa. "A simple and efficient hybrid maximum power point
paper. This paper focuses on improving the efficiency of PV tracking method for PV systems under partially shaded condition" ,
panels. This algorithm successfully tracks the global IECON 2013 - 39th Annual Conference of the IEEE Industrial
maximum value unlike the conventional algorithms which are Electronics Society, 2013.
used for MPPT. The control design is also made simpler for
[9]. Ahmed M. Atallah, Almoataz Y. Abdelaziz, and Raihan S. Jumaah
MPPT. The simulation results are also checked with that of the “Implementation pf Perturb and Observe MPPT of PV system with
hardware results and the results are satisfactory, which shows direct control method using buck and buck-boost converters” Emerging
the efficiency of ACO algorithm to get maximum power. The Trends in Electrical, Electronics & Instrumentation Engineering: An
algorithm also has higher convergence rate and lesser number international Journal (EEIEJ), Vol. 1, No. 1, February 2014
of iterations are required to get the result. Thus this algorithm [10]. Ankit Digrawal “Simulation Study of Photovoltaic System with MPPT
is more advantageous compared to other algorithms. Algorithms” International Journal of Science and Research (IJSR)
ACKNOWLEDGMENT
The authors gratefully acknowledge support from
Department of Science and Technology (DST), Government of
India, Project No. DST/TSG/NTS/2013/59 and VIT University,
Vellore, India.
