Short-term Photovoltaic Power Prediction Based on Moderate-resolution Imaging Spectroradiometer Clear Sky Data

In order to predict photovoltaic power effectively, a method based on improved Bird model and FPGA fast neural network is proposed in article. This method forecast the thickness of Aerosol Optical (Aerosol Optical the Depth, AOD) through FPGA (Field Programmable Gate Array) constructed neural network first, then ameliorate BIRD model. We used AOD and meteorological data to predict different period of time's solar radiation intensity through BIRD model in clear sky, corrected cloud model with fuzzy variable, built irradiance forecast model and then calculated the photovoltaic panels outputs with reference of specifications. The result shows that the proposed method based on MODIS has high accuracy and it is a practical significance to a certain degree.