Stochastic Optimization of Pricing Strategy for Electricity Retailers Based on One-Leader-Many-Follower Game

To improve the revenue of electricity retailers, a stochastic optimization pricing strategy considering the uncertainty of photovoltaic-storage users is proposed. Firstly, considering the initiative and decision-making ability of the retailer-load-storage triad, with the retailer as the leader and the load and storage as followers, a one-leader-many-follower game model between the retailer and multiple photovoltaic-storage users is established. Secondly, considering the uncertainty of user-side load forecasting and photovoltaic output, typical scenarios are generated based on the cumulative distribution function clustering of prediction error. Finally, a neural network agent model is used to fit the purchasing and selling behavior of photovoltaic-storage users to electricity retailers, replacing the decision-making process of storage output and the demand response behavior of loads. The genetic algorithm is used for iterative optimization and the neural network agent model is corrected, continuously updating the internal pricing of the retailer. The effectiveness of the proposed model and algorithm is verified through a test system containing four photovoltaic-storage users, increasing the expected revenue of retailers considering the impact of uncertainty factors, and avoiding the complex invocation of lower-level optimization models, saving computation time.