Prediction of Photovoltaic Power Generation Based on General Regression and Back Propagation Neural Network

摘要： Based on the general regression (GR) and back propagation (BP) neural network prediction method, this work forecasts the generated power of photovoltaic (PV) power station. First, the Pearson correlation coefficient method was used to analyze the meteorological factors. The degree of correlation between complex weather factors and PV power output was differentiated and the irradiance and battery temperature were selected as the important influencing variables. Second, the weather was classified according to the certain classification criteria. Then, we established the model by using GR and BP neural network prediction methods. The relative errors were within acceptable limits. The former model is more convenient while the latter model has better nonlinear fitting capacity. The results of the two models are compared and analyzed. We find out that the BP neural prediction method have better prediction results than GR method on PV power generation. Our findings can not only provide valuable information for the optimal dispatching of micro-grid and photovoltaic power, but also be of great significance in energy management and hierarchical control of micro-grid.