Ultra-short-term Photovoltaic Power Prediction Based on Elman Neural Network and Satellite Cloud Images

Solar photovoltaic (PV) power generation has developed rapidly in recent years due to its clean and pollution-free characteristics. However, the variability of weather makes PV power generation very volatile and random, which brings challenges to the planning and operation of the existing power system. Therefore, accurate prediction of PV power generation is of great significance to improving the reliability of power system operation. This paper proposes a PV power prediction model based on Elman neural network and gray value extraction method of satellite cloud images, which effectively utilizes weather pattern and improves the accuracy of PV power prediction. Firstly, the satellite cloud image is grayed to get the gray value, and the gray value interval is clustered. Then the clustering result is used as the input of Elman neural network to build the PV power prediction model. Finally, the actual data are used to analyze the calculation example. The simulation results show that the optimization model after adding the gray value information of the satellite cloud image is more accurate than the original model. The relative coefficient of variation (RCV) is reduced from 5.42 % to 4.01 %, and the accuracy is increased by 1.41%. It provides a reference for using satellite cloud images to predict PV power generation.