Photovoltaic Generation Forecasting Based on Spatial Correlation and Installed Capacity Characteristics

In view of the problem that a large number of distributed photovoltaic output data cannot be obtained and the impact of constant changes in photovoltaic installed capacity on prediction accuracy in photovoltaic output prediction, this paper proposes a photovoltaic output prediction method based on spatial correlation and installed capacity characteristics. First, the correlation of photovoltaic output within a certain spatial range was analyzed, and the centralized photovoltaic power station with high correlation was selected as the sample power station; secondly, the output and capacity data in different weather were determined according to the clustering method, the equivalent capacity of each photovoltaic power station relative to the sample power station was obtained, and the output and equivalent capacity of each sample power station were linearly fitted; finally, the output and equivalent capacity model were used to predict the output of the sample power station, and the distributed photovoltaic output was predicted by using the CNN-LSTM model, and the output and equivalent capacity model were combined to predict the distributed photovoltaic output. The prediction results show that this method can predict distributed photovoltaic output more accurately, and can overcome the problems of change in photovoltaic installed capacity.