Abstract
This paper proposes a multiagent based bi-level operation framework for thelow-carbon demand management in distribution networks considering the carbonemission allowance on the demand side. In the upper level, the aggregate loadagents optimize the control signals for various types of loads to maximize theprofits; in the lower level, the distribution network operator makes optimaldispatching decisions to minimize the operational costs and calculates thedistribution locational marginal price and carbon intensity. The distributedflexible load agent has only incomplete information of the distribution networkand cooperates with other agents using networked communication. Finally, theproblem is formulated into a networked multi-agent constrained Markov decisionprocess, which is solved using a safe reinforcement learning algorithm calledconsensus multi-agent constrained policy optimization considering the carbonemission allowance for each agent. Case studies with the IEEE 33-bus and123-bus distribution network systems demonstrate the effectiveness of theproposed approach, in terms of satisfying the carbon emission constraint ondemand side, ensuring the safe operation of the distribution network andpreserving privacy of both sides.