P2P Energy Trading in a Smart Residential Environment with User Behavioral Modeling

Peer-to-peer (P2P) energy trading is a decentralized energy market for consumers, with or without energy generation capabilities, to trade energy among each other. By not considering user behavior and making unrealistic assumptions about their participation and compliance, the effectiveness of the existing P2P energy trading approaches has been limited in literature. To over-come these limitations, in this work, we propose an automated P2P energy trading framework that incorporates user behavioral modeling into the problem while also learning the optimal trading strategies and parameters online. We devise mechanisms to match energy production and demand that allocates energy between sellers and buyers. We also propose reinforcement learning-based automated pricing solutions to improve sellers' long-term profit. Testing proposed framework with real traces of energy consumption and production, and online learning, 26% higher perceived value was observed for buyers with 7% more reward for sellers compared to state-of-the-art approaches.