Data-driven Boundary Optimal Control for Diffusion-Reaction Processes Considering Actuator Dynamic Properties

We studies a data-driven optimal boundary control method for diffusion-reaction processes coupled with actuator dynamics. Considering the actuator characteristics as ordinary differential equations, we first rewrite the coupled distributed parameter system with boundary control as one with distributed control inputs. Subsequently, applying Bellman optimality principle, a novel policy iterative optimal control algorithm is designed, which collects state datasets along the system trajectory and then iteratively computes the optimal control policy offline. There is no need to run the system online and the collected dataset can be reused, greatly saving computational resources.