Interactive Learning - Implementation of ChatGPT and Reinforcement Learning in Local Energy Trading

Within the Local Electricity Market (LEM), Distributed Energy Resources (DERs) are permitted to trade their excess energy to others under the scheme of peer-to-peer (P2P) energy trading. During the trading procedure, DERs in the prototype are required to participate in each trading activities across various time slots, which can be a pain point for real human beings. To address this, AI technologies like Reinforcement Learning (RL) are introduced to alleviate this pain, aiming to provide end users with more free time. In typical RL methods, the system learns from end users historical data to simulate their trading behaviour, known as auto trading. However, unforeseen situations not covered in historical data may arise. In this paper, the author proposes an interactive learning scheme based on ChatGPT and Reinforcement Learning to support both auto trading and interactive trading simultaneously.