Power Prediction Method of Distributed Photovoltaic Digital Twin System Based on GA-BP

Abstract

Aiming at the problems of low accuracy, long time and large error of traditional distributed photovoltaics (D-PV) power prediction methods, a power prediction method of D-PV digital twin system based on GA-BP-NN is proposed. Firstly, by analyzing the basic principles and key features of the power system digital twin system, the digital twin structure system of the photovoltaic power generation power prediction system is constructed. Then, through the quantitative analysis of photovoltaic output fluctuation and photovoltaic operating status, a photovoltaic power generation power prediction model is built in the digital twin system based on the GA-BP-NN algorithm. Finally, the proposed D-PV power prediction method and the other two methods are compared and analyzed under the same conditions through simulation experiments. The results show that the power prediction accuracy and time consumption of the method proposed in this paper are the best in three different meteorological environments, the highest accuracy is 95.24%, and the minimum time consumption is 6.53s, and the algorithm performance is better than the other three comparisons algorithm.