A multi-scale energy storage configuration planning method with improved MMD-GAN wind–photovoltaic scene generation

Abstract

Increasing proportion of new energy puts new demands on the flexibility regulation capability of the new power system. To this end, a long–short term wind–PV–storage station cooperative optimization planning method is established to take into account the economy and multi-scale and multi-objective. Firstly, an improved MMD-GAN is used to generate the wind and solar output scene. Month labels are added to depict the power output accurately, and multi-scenes are assigned by boundary-divided cluster method. Secondly, based on the power market clearing and different energy storage operation characteristics, a two-layer optimization model with long–short time scale is established. The upper layer determines the energy storage configuration, and the lower layer simulates the operation strategy. Through the information interaction between the upper and lower layers, the energy storage parameters are dynamically optimized in the production simulation, to realize the economically optimal power-capacity configuration scheme. Finally, the effectiveness of the proposed method is verified on the actual data of a wind–PV-storage power plant in Gansu. The results show that the improved MMD-GAN method reduces over 45% on the Wasserstein distance than the comparison methods, which can improve the accuracy of the generated scenarios. The combined BES-HES configuration reduces the wind–PV abandonment rate by more than 3.22% compared to the single storage configuration. Based on the volatility LMP strategy, it takes advantage of the price differences between different time periods to expand arbitrage space and achieve economic optimization.