Research on Maximum Power Point Tracking for Photovoltaic Systems based Artificial Neural Network Algorithm

• Main Authors: Chin-chang Hsu, 許金長
• Other Authors: Yi-Hua Liu
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/w9pfne

碩士 === 國立臺灣科技大學 === 電機工程系 === 99 === Low power PV systems are commonly used in stand-alone applications such as LED lightings. For these systems, a simple and cost-effective maximum power point tracking (MPPT) solution is essential.For low power PV systems, analog MPPT techniques such as two-line approximation (TLA) method and curve approximation (CA) method are preferred due to their simple structure, low cost, fast tracking speed and high tracking efficiency.However, these methods exploit the relationship between the values of panel voltage and current, which often should be obtained by complicated numerical simulations or experiments. To assist the system designer and simplify the design procedure, an artificial neural network (ANN) based method is proposed in this thesis to obtain the required parameters of the TLA MPPT method and CA MPPT method directly. The input variables are the nominal open-circuit voltage (VOC,STC), the nominal short-circuit current (ISC,STC), the voltage at the MPP (VMP), the current at the MPP (IMP) and the cell number (NS) in the utilized PV array. These data can easily be obtained from datasheets of PV panels. After collecting all the PV panel datasheets of the top 10 PV panel manufacturers in 2010, simulations are made to obtain the parameters of the TLA and CA methods. These data are then utilized as the training data of the proposed ANN. Simulation results show that the proposed ANN can precisely predict the parameters of the TLA and CA methods, with largest error smaller than 10%.