Research on the distributed photovoltaic power prediction method based on CNN–LSTM

Abstract The proposed model is an enhanced prediction model that utilizes a convolutional neural network (CNN) and a long- and short-term memory network (LSTM). The model first preprocesses the raw data to determine the key input features of the prediction model. Subsequently, a hybrid model comprising a CNN and a LSTM network was developed. The CNN is responsible for capturing spatial correlations between different geographical locations, while the LSTM focuses on identifying long-term dependencies in the photovoltaic time series. The experimental results demonstrate that the CNN–LSTM-based prediction model attains high prediction accuracy, thereby substantiating the efficacy and preeminence of this methodology.