Short-Term Photovoltaic Power Prediction Based on 3DCNN and CLSTM Hybrid Model

Abstract

The power prediction of photovoltaic (PV) generation is an important basis for the power system to formulate power generation plans and coordinate dispatch. However, due to the randomness and intermittency of the PV generation process, there is still much room for improvement in the accuracy of PV power prediction. This paper proposes a PV power prediction method based on a mixed model of three-dimensional convolutional neural network (3DCNN) and convolutional long short-term memory network (CLSTM). The input parameters of the model are determined using the correlation coefficient method, and the accuracy of the prediction model is evaluated using three indicators: root mean square error, mean absolute error, and mean absolute percentage error. To verify the applicability and correctness of the model, the prediction method based on this mixed model is ap-plied to the output prediction of a certain PV power station in Shandong, China. The PV output power under different weather conditions with the same input sequence and under different input sequences is predicted, and the results show that the prediction effect based on the 3DCNN+CLSTM mixed model is better than that of the 3DCNN model, the CLSTM model, and the BP neural network model under both scenarios.