Control-Informed Reinforcement Learning for Chemical Processes

TL;DR

This work addresses the challenge of achieving robust, sample-efficient control for nonlinear processes by fusing classical PID control with deep reinforcement learning. The authors introduce CIRL, a framework that embeds a PID controller layer within a deep RL policy, enabling learned adaptive PID gains while preserving the stabilizing and interpretability benefits of PID control. Through a nonlinear CSTR case study, CIRL demonstrates improved setpoint tracking, generalization beyond training conditions, and enhanced disturbance rejection compared with static PID and model-free RL, with the added benefit of faster, safer learning due to the embedded prior control structure. The approach is agnostic to the underlying policy optimization method and leverages evolutionary strategies to optimize the policy, offering a practically viable path toward robust, industrial-scale RL-based control implementations.

Abstract

This work proposes a control-informed reinforcement learning (CIRL) framework that integrates proportional-integral-derivative (PID) control components into the architecture of deep reinforcement learning (RL) policies. The proposed approach augments deep RL agents with a PID controller layer, incorporating prior knowledge from control theory into the learning process. CIRL improves performance and robustness by combining the best of both worlds: the disturbance-rejection and setpoint-tracking capabilities of PID control and the nonlinear modeling capacity of deep RL. Simulation studies conducted on a continuously stirred tank reactor system demonstrate the improved performance of CIRL compared to both conventional model-free deep RL and static PID controllers. CIRL exhibits better setpoint-tracking ability, particularly when generalizing to trajectories outside the training distribution, suggesting enhanced generalization capabilities. Furthermore, the embedded prior control knowledge within the CIRL policy improves its robustness to unobserved system disturbances. The control-informed RL framework combines the strengths of classical control and reinforcement learning to develop sample-efficient and robust deep reinforcement learning algorithms, with potential applications in complex industrial systems.

Figures (14)

• Figure 1: The RL framework
• Figure 2: Deep policy network $\pi_\theta$
• Figure 3: Deep Q-function $Q_\phi$
• Figure 4: CIRL Agent
• Figure 5: Block diagram of the policy optimization algorithm