New MPPT Controller Design for PV Arrays Using Neural Networks

New MPPT Controller Design for PV Arrays Using Neural Networks (Zanjan City Case Study) Mehran Habibi and Alireza Yazdizadeh Dep. of Electrical Engineering, Power and Water University of Technology, P. O. Box 16765-1719, Tehran, Iran

Abstract. This paper proposes a novel Voltage-Based Maximum Power Point Tracking (MPPT) technique by introducing a new and simple tracking algorithm. Compared with other Voltage-Based MPPT methods which assume the optimal voltage factor as a constant parameter, in the proposed algorithm, the optimal voltage factor is instantaneously determined by a neural network. The proposed MPPT algorithm is applied to a Buck regulator to regulate the output power at its maximum possible value. Simulation results show the excellent MPPT performance in different temperatures and insulation levels during a day in a specific area. Keywords: Maximum power point tracking, Photovoltaic, Neural networks, Buck converter.

1 Introduction The renewable energy sources are attracting more attentions in recent decades. Among them, the solar energy has some important aspects that discriminate it from other sources. It is clean, pollution free, inexhaustible and it has a secure power source. In spite of these attractive features, the photovoltaic generation is an expensive and low energy conversion efficiency system compared to other electric power sources. The main reason of the low electrical efficiency of a PV array is the nonlinear I-V and P-V characteristics of PV arrays and the effects of environmental conditions on these characteristics such as: varying temperature and different insulation levels. Among all the increasing output techniques, “Look-Up Table” method and “Computational” methods are two main categories of the so far introduced methods. In the “Look-Up Table” method [6] the drawback is impossibility of storing all the system conditions because of the time varying and nonlinear nature of the solar cells and their great dependency on irradiation and temperature levels [7]. In “Computational” method the I-V characteristics of solar panel is modeled by using mathematical equations and based on this model, the maximum power point is obtained as a function of the PV short circuit current [8] or PV open circuit voltage [6-8]. Some researchers have used neural networks in order to model the PV characteristics for estimating maximum power point operating conditions [3,4], while others have used neural networks in the feedback control loop of the control system [2,5]. In some research works the open circuit voltage is used for MPPT [6,7]. W. Yu, H. He, and N. Zhang (Eds.): ISNN 2009, Part II, LNCS 5552, pp. 1050–1058, 2009. © Springer-Verlag Berlin Heidelberg 2009

New MPPT Controller Design for PV Arrays

1051

This paper, proposes a new Voltage-Based MPPT algorithm by introducing the optimal voltage factor, and tries to use the salient features of neural networks to estimate the nonlinear curves of maximum power points and optimal voltage factors in different temperatures and insulation conditions. In this method, a neural network estimates all the maximum power points only by a limited amount of maximum power points and by saving only some conditions the other conditions are accessible. Therefore, by using a neural network, the difficulties of the “Look-Up Table” method is solved. The simulation results show satisfactory performance of the controller and its effects on increasing the output power efficiency.

2 Solar Cell Mathematical Model The I-V characteristics of a PV cell by using the equivalent circuit of a solar cell (Fig. 1) can be expressed as:

I = IL −

⎛ ⎛V + R I s ⎜ ⎜ ⎜ ⎜ nkT / q ⎝ IO ⎜ e ⎜ ⎜⎜ ⎝

⎞ ⎞ ⎟ ⎟ ⎟ ⎟ ⎠ −1 ⎟ ⎟ ⎟⎟ ⎠

−

V + Rs I R sh

Fig. 1. Equivalent circuit of a solar cell

where:

V : the output voltage of the solar cell. I : the output current of the solar cell. I L : the photo current (representing insulation level).

I O : the reverse saturation current. T : the temperature of the solar cell. Rs : series cell resistance. Rsh : shunt cell resistance. q : electron charge. k : Boltzman constant. n : ideality factor (1