Hierarchical RL-MPC for Demand Response Scheduling

TL;DR

This paper introduces a hierarchical RL-MPC framework for demand response in air separation units, combining reinforcement learning with model predictive control to improve efficiency, constraint satisfaction, and robustness in operational management.

Contribution

The paper proposes a novel hierarchical RL-LMPC framework that enhances demand response scheduling by integrating data-driven RL with traditional control, improving sample efficiency and constraint handling.

Findings

Improved sample efficiency during training.

Better constraint satisfaction compared to direct RL.

Achieves comparable economic performance with increased robustness.

Abstract

This paper presents a hierarchical framework for demand response optimization in air separation units (ASUs) that combines reinforcement learning (RL) with linear model predictive control (LMPC). We investigate two control architectures: a direct RL approach and a control-informed methodology where an RL agent provides setpoints to a lower-level LMPC. The proposed RL-LMPC framework demonstrates improved sample efficiency during training and better constraint satisfaction compared to direct RL control. Using an industrial ASU case study, we show that our approach successfully manages operational constraints while optimizing electricity costs under time-varying pricing. Results indicate that the RL-LMPC architecture achieves comparable economic performance to direct RL while providing better robustness and requiring fewer training samples to converge. The framework offers a practical solution for implementing flexible operation strategies in process industries, bridging the gap between data-driven methods and traditional control approaches.