Pricing Strategy for Regional Integrated Energy System Considering Privacy Based on Deep Reinforcement Learning

With deregulation of the energy market, the pricing strategy of energy sellers in a regional integrated energy system (RIES) can affect the interests of all participants in the market and the operation of the system. This paper proposes a pricing strategy for integrated energy service providers in RIES based on a deep reinforcement learning (DRL) algorithm considering privacy protection. The transaction process between the integrated energy service provider (IESP) and user aggregators (UAs) in RIES is modeled as a Stackelberg game. IESP serves as the leader in making retail prices, and different UAs serve as followers in optimizing their energy consumption strategies. Considering UAs' strategies are temporally coupled, a Markov decision process (MDP) is designed differently from existing studies. Case studies demonstrate that the proposed method is accurate and stable when solving a Stackelberg equilibrium without privacy leakage. The obtained pricing strategy avoids unreasonable pricing and guarantees the revenue of IESP and the energy demand of UAs.