Community Energy Management Using MARL: Synergy of Price-Based and Incentive-Based Demand Response

The decentralization trend of the power system highlights the role of consumers in future plans. Mediatory infrastructures, such as the energy community, play a vital role in the realization of customer-centric behaviors like Demand Response (DR). This paper proposes a two-stage DR strategy within a residential energy community, integrating Price-Based DR (PBDR) and Incentive-Based DR (IBDR). Multi-agent reinforcement Learning (MARL) is utilized for optimal policy training of the community aggregator and Home Energy Management System (HMES) of the consumers. The PBDR, as the initial stage, involves dynamic pricing of the aggregator to influence consumers' behavior. This leads to a significant reduction in peak load and smoothing of the load pattern by reducing/ shifting the consumption in/ from high-priced timeslots. The PBDR program alone may be insufficient for meeting consumption thresholds imposed by the upstream operator. Therefore, an IBDR program is introduced as the second stage, offering incentives to consumers to reduce their demands further. The synergy of these two programs can successfully modify the community demand. Simulation studies show the improvements in load characteristics of the community and a positive overall reward for both aggregator and consumers, suggesting a proper balance of objectives and efficient energy management in the energy community.