Solar Radiation Forecasting based on Neural Network in Guangzhou

Solar power generation occupies a leading position in new energy generation technology, however output power fluctuations caused by the uncertainty of solar radiation changes may seriously affect the safety and stability of the power grid. Therefore, solar radiation data is the main basis for evaluating the output of photovoltaic (PV) power generation. Considering the influence of various meteorological factors on the intensity of solar radiation, accurate prediction of solar radiation intensity is particularly important in the development of the new energy industry. In this paper, we first analyze the relationship between various meteorological factors and solar radiation intensity; secondly, we built a BP neural network to construct a model; then we proposed to use rainfall, humidity, and clear-air index as inputs to the solar radiation intensity prediction system. We verified that the predicted and actual output values of the model have a root mean square error (RMSE) of less than 0.5. Finally, the training and testing samples of different months and regions were selected to test the model which verified the universal applicability of the proposed model.