Multi-Microgrids Peer to Peer Electricity-Carbon Joint Trading Method Based on Multi-Agent Ethical Enhancement Technology

This is the authors' abstract. We don't add key points for this paper.

Electricity-carbon joint energy management (ECCEM) for multi-microgrids (MGs) has great potential to promote renewable energy accommodation, balance the benefits of MGs, and reduce carbon emissions. This article proposes a multi-agent safe reinforcement learning algorithm incorporating ethical principles for ECCEM of MGs to achieve low-carbon economic operation of MGs while ensuring adherence to human ethical values. First, to reduce carbon emissions, carbon trading is integrated into the P2P trading model, and a novel joint electricity-carbon trading framework is proposed. Then, a safe deep reinforcement learning method based on the upper confidence bound and augmented lagrangian framework was designed to address the physical security of electricity power balancing while aligning with ethical principles of do-no-harm. Finally, a privacy protection module based on equidistant geometric transformations has been designed to ensure data privacy during the multi-participant bidding process in the P2P market, adhering to ethical principles of privacy. Case studies involving a four-MGs demonstrate that the proposed method achieves not only the primary goals of reducing carbon emissions, improving economics, and promoting renewable energy accommodation in the process of MGs electricity-carbon trading, but also achieves the ethical goals of do-no-harm and privacy protection.