A RSCMAC Based Forecasting for Solar radiance with Meteorological Data and Wavelet Analysis

• Main Authors: Hsing-Yang Chou, 周幸洋
• Other Authors: 江青瓒
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/19678682562338028399

碩士 === 健行科技大學 === 電子工程系碩士班 === 103 === In recent years, due to the impact of global warming, many countries pay attention to renewable energy. Under the government strongly promotion, total installed capacity of renewable energy generation systems is increased year by year. Among the renewable energy sources, PV( Photovoltaic ) system is one of the most important energy. Because of the power generation of PV systems changes as the solar irradiance varies, its power generation is not stable and affects overall electricity grid configuration, scheduling and stability. Therefore, accurately predict PV system power generation is very important for regional power dispatch. Solar irradiance is the most important factor to effect PV system power generation. This study focuses on solar irradiance short-term prediction; use RSCMAC (Recurrent Simple addressing structure for Cerebellar Model Articulation Controller) as the basis to combine meteorological data and wavelet analysis to establish a solar irradiance prediction model. Furthermore, this study uses the peak value and the amount of change of solar irradiance to establish standard wave training patterns to be used in solar irradiance prediction model, this can solved the training and testing deviation caused by weather forecast error, training wave not generalized, etc. Excellent performance is obtained from simulation test results.