A Collaborative Machine Learning Based Decisionmaking Method for Carbon Electricity Trading in Multi-Sector Power Union

This paper explores a reinforcement learning-based model of a tripartite cooperative electricity market that aims to optimise electricity trading and revenue distribution among industrial, commercial and residential users. By introducing the Renewable Energy Certificates (RECs) trading mechanism, the article analyses the dynamic trading strategies and their interactions among different user types in the electricity market. The paper proposes an optimisation approach based on the MultiAgent Soft Actor-Critic (MASAC) algorithm to cope with the heterogeneity of participants' objectives and complex multiparty cooperation in electricity markets. The model incorporates time-of-use tariffs and real-Time grid demand-supply conditions to develop an adaptive pricing platform that fosters cooperative behaviour and enhances energy efficiency. Coordinating internal pricing and transaction scheduling supports more effective market operations. The study further examines how the model's dynamic adjustment strategies and collaborative mechanisms contribute to improved economic performance and system stability within a real-Time market context.