Study on photovoltaic power prediction with TSO-LSTM-XGBoost coupled model accounting for weather factors

Abstract

ABSTRACT The explicit prediction of PV power itself is of great significance to the scheduling and operation of the power grid. To ensure the stable operation of the power system, this paper proposes a coupled model for PV power prediction based on TSO-LSTM-XGBoost. This model considers the weather factor using the tuna swarm algorithm(TSO) to optimize the long and short-term memory network model(LSTM) to overcome the shortcomings of blindness and time-consuming in the process of randomly selecting the parameters of the LSTM model. At the same time, using the extreme gradient boosting model (XGBoost), the algorithm is improved and corrected for the large prediction error in cloudy and rainy weather, and the weighted error method is used to couple the model to obtain the final prediction results. Finally, the accuracy of the proposed model is verified by comparing the PV system and meteorological data of a certain region in Shenzhen, China. The results show that the proposed TSO-LSTM-XGBoost coupled model has a value of 71.98 for MAE in cloudy days and 82.09 for MAE in rainy days, and the prediction accuracy is better than PSO-LSTM and LSTM.