Projects

IIT Kharagpur Foreign Training Program 2021-22

Multi-Agent RL for demand-response management

We consider a residential microgrid (see figure), which consists of residential households/buildings connected to the distribution grid, via the point of common coupling (abbreviated as PoCC). Each household has a (net)demand owing to the uncontrollable loads (in the form of household appliances) and/or an uncontrollable generation (in the form of rooftop PV panels). The households share a community battery energy storage system (abbreviated as CBESS) which is collectively owned by all the households in the residential microgrid and acts as the only flexible/controllable load. The households take part in the community demand response program, by exploiting the flexibility of the CBESS in response to the “dynamic” electricity tariff of the energy retailer. The electricity tariff is “dynamic” in nature as it is influenced by the aggregate electricity exchange of the residential microgrid at the point of common coupling, which indirectly depends on the charging/discharging of the CBESS. In this study, we assume that all the households are cooperative in nature and are willing to coordinate with each other so as to control the CBESS.

Useful reading

On Multi-agent RL

Buşoniu, Lucian, Robert Babuška, and Bart De Schutter. "Multi-agent reinforcement learning: An overview." Innovations in multi-agent systems and applications-1 (2010): 183-221.

On RL

Sutton, Richard S., and Andrew G. Barto. Reinforcement learning: An introduction. MIT press, 2018.

On Demand response

Pinson, Pierre, and Henrik Madsen. "Benefits and challenges of electrical demand response: A critical review." Renewable and Sustainable Energy Reviews 39 (2014): 686-699.