A Centralised Soft Actor Critic Deep Reinforcement Learning Approach to District Demand Side Management through CityLearn

TL;DR

This paper demonstrates a centralized Soft Actor Critic deep reinforcement learning controller for district demand side management, achieving high performance in the CityLearn Challenge and handling diverse building types and climates.

Contribution

It introduces a novel centralized DRL approach using Soft Actor Critic for district demand management, showcasing its effectiveness across varied building and climate conditions.

Findings

Achieved an average score of 0.967 in the CityLearn Challenge.

Secured second place in the challenge.

Proved adaptability to different climates and building types.

Abstract

Reinforcement learning is a promising model-free and adaptive controller for demand side management, as part of the future smart grid, at the district level. This paper presents the results of the algorithm that was submitted for the CityLearn Challenge, which was hosted in early 2020 with the aim of designing and tuning a reinforcement learning agent to flatten and smooth the aggregated curve of electrical demand of a district of diverse buildings. The proposed solution secured second place in the challenge using a centralised 'Soft Actor Critic' deep reinforcement learning agent that was able to handle continuous action spaces. The controller was able to achieve an averaged score of 0.967 on the challenge dataset comprising of different buildings and climates. This highlights the potential application of deep reinforcement learning as a plug-and-play style controller, that is capable of handling different climates and a heterogenous building stock, for district demand side management of buildings.