Photovoltaic Power Forecasting And Its Application In Micro Grid

With Governments worldwide advocating the new energy policy,the share of the global power production coming from solar power is increasing.However the intermittency and volatility of solar power become a huge challenge for electric power systems at the higher penetration levels.Photovoltaic power forecasting is a key technology to enable an efficient management and control of the photovoltaic system.This thesis firstly introduces the research status and developing direction of the photovoltaic power forecasting technology,and the basic principles and output characteristics of the photovoltaic power generation are briefly described.Then a thorough study of correlation between the photovoltaic power and meteorological factors is carried out.To achieve efficient and precise forecast of the photovoltaic power,this thesis analyzes and compares solar radiation models in existence,and proposes a new solar radiation cloudy model to realize the accurate forecast of solar radiation next day.After obtaining the forecast value of next day solar irradiance,a photovoltaic power forecasting model based on BP neural networks optimized by Genetic Algorithms and a photovoltaic power forecasting model based on GRNN(Generalized Regression Neural Networks)are designed and implemented.The practical application of the photovoltaic power forecasting system in micro grid is introduced.The data of historical power and meteorology is used to train proposed forecasting models,and the forecasting performance of proposed forecasting models is evaluated.Experiments show these two models have high accuracy and reliability,and the practical needs are satisfied.