Breadcrumb
- Home
- Publications
- Proceedings
- 2025 AIChE Annual Meeting
- Education Division
- Curriculum Assessment and Core ChE Courses
- (532g) Integrating Reinforcement Learning into Process Control Curriculum with a Water Level Control Example
We developed a Simulink-based water level control example using reinforcement learning that instructors can easily integrate into their curriculum. Simulink is a widely used block-based environment that allows students to model dynamic systems, design control algorithms, and run closed-loop simulations. Many chemical engineering process control courses utilize Simulink.
Our example consists of a Simulink model of a water tank system controlled with a PI controller to maintain the water level at a desired set point. To implement reinforcement learning approach, we created an environment and replaced the PI controller with an Actor-Critic Reinforcement Learning Agent. We used a Deep Deterministic Policy Gradient (DDPG) agent as the actor, trained it, and validated the results. The reinforcement learning approach was effective at maintaining the water level at the set point. With both PI control and reinforcement learning components in this example, instructors can easily build on the existing syllabus and introduce reinforcement learning using a system that students are already familiar with.
To further assist instructors in integrating reinforcement learning into their chemical engineering curriculum, we will provide freely available self-paced online training courses, videos, and project ideas.