Dynamic Pricing Strategy in Electricity Trading Market Based on Reinforcement Learning

Chunlin Hu, Dou An

Abstract: Intermediaries play an important role in electricity trading in the smart grid. However, smart grids are highly dynamic and complex due to diverse consumer demand, uncertain electricity market prices, and competition among different intermediaries. It brings a huge challenge for intermediaries to set reasonable electricity prices to maintain supply and demand balance and maximize their profits. This paper presents an electric energy trading model based on an intermediary. In this model, the electricity purchased by the user is negatively correlated with the retail price published by the intermediary. Intermediaries can make maximum profit by adjusting retail prices. Considering that deep reinforcement learning (DRL) can solve uncertain decision-making problems, this paper uses the deep Q-network (DQN) algorithm to develop real-time pricing strategies for intermediaries in the electricity market. Then, the validity of the pricing strategy of the intermediary is verified by the simulation of electricity consumption data in the real world. After comparison with non-learned fixed pricing strategies and random pricing strategies, the simulation results show that the price strategy based on DRL is more profitable, which further explains the effectiveness and superiority of the price strategy based on DRL.