Why Reinforcement Learning in Energy Systems Needs Explanations

TL;DR

The paper addresses the trust gap in applying reinforcement learning to energy systems with distributed energy resources by arguing for explainability as essential for deployment. It surveys the state of explainable AI in energy and proposes a focused effort on explainable RL for small-scale buildings with PV and energy storage, using methods such as Deep Q-Networks and Proximal Policy Optimization, complemented by break-down profile explanations and counterfactuals. The planned contributions include assessing RL applicability to energy optimization, comparing RL methods, and developing interpretable RL frameworks to enhance trust and facilitate large-scale adoption. This work aims to enable transparent, robust, and scalable RL-based energy management, supporting broader integration of renewables and demand-side strategies in real-world grids.

Abstract

With economic development, the complexity of infrastructure has increased drastically. Similarly, with the shift from fossil fuels to renewable sources of energy, there is a dire need for such systems that not only predict and forecast with accuracy but also help in understanding the process of predictions. Artificial intelligence and machine learning techniques have helped in finding out wellperforming solutions to different problems in the energy sector. However, the usage of state-of-the-art techniques like reinforcement learning is not surprisingly convincing. This paper discusses the application of reinforcement techniques in energy systems and how explanations of these models can be helpful