A Neural Network-Based Rapid Maximum Power Point Tracking Method for Photovoltaic Systems in Partial Shading Conditions
- PDF / 1,376,888 Bytes
- 11 Pages / 612 x 792 pts (letter) Page_size
- 56 Downloads / 204 Views
CT CONVERSION OF SOLAR ENERGY INTO ELECTRICAL ENERGY
A Neural Network-Based Rapid Maximum Power Point Tracking Method for Photovoltaic Systems in Partial Shading Conditions Adi Kurniawana, b, * and Eiji Shintakub aDepartment bDepartment
of Marine Engineering, Institut Teknologi Sepuluh Nopember, Surabaya, Indonesia of Transportation and Environmental Systems, Hiroshima University, Hiroshima, Japan *e-mail: [email protected] Received February 19, 2019; revised June 6, 2019; accepted January 7, 2020
Abstract—The maximum power point tracking (MPPT) controller holds an important role in increasing the efficiency of the photovoltaic (PV) system. However, conventional MPPT techniques may fail to locate the global maximum power point (GMPP) under partial shading conditions (PSC). Hence, to optimize the efficiency of PV systems, we introduce a new MPPT technique for PSC. The proposed method employs an artificial neural network (ANN) to predict the area of the GMPP, and the classic perturb and observe (P&O) algorithm to locate the exact position of the GMPP. We validated the effectiveness of the technique using computer simulations performed with the MATLAB/Simulink program, the results of which verified that it can track the GMPP faster than other methods. Keywords: artificial neural network, computer simulation, global maximum power point, hybrid method, partial shading conditions, perturb and observe DOI: 10.3103/S0003701X20030068
INTRODUCTION A range of maximum power point tracking (MPPT) algorithms have been used in photovoltaic (PV) systems. Due to their simplicity, the perturb and observe (P&O) and incremental conductance methods are most often used in commercial products [1]. However, faster and more accurate MPPT techniques have been developed, including the numerical method [2, 3], extremum seeking control [4], and artificial intelligence (AI) algorithms, such as fuzzy logic [5], artificial neural network (ANN) [6, 7], neuro-fuzzy combinations [8, 9] and particle swarm optimization (PSO) techniques [10]. Because the MPPT algorithms are designed for uniformly irradiated PV panels in a string system, they may fail to locate the global maximum power point (GMPP) during partial shading conditions (PSC) due to the appearance of several local maximum power points (LMPP), decreasing the total efficiency of the system. The failure of the P&O MPPT algorithm to find the GMPP during PSC has been proven under computer simulation in [11]. Figure 1 illustrates the output of a PV string consisting of three panels connected in a series which absorbs three levels of solar radiation due to partial shading. Two possible solutions for addressing PSC are the use of a distributed MPPT (DMPPT) [11] or specifi-
cally designed MPPT algorithms for PSC. The MPPTs for PSC can be classified into three categories: soft computing, segmental search, and hybrid methods. Hybrid methods are preferable to develop as they are gen erally easier to implement than soft computing methods, and are more accurate than segmental search methods [12]. Hyb
Data Loading...
