Resource-aware model predictive control of spatially distributed processes using event-triggered communication

This work focuses on the design of a resource-aware model predictive control (MPC) system for spatially distributed processes with low-order dynamics and a limited number of output measurements. A reduced-order model that captures the dominant process dynamics is initially obtained and used to design a Lyapunov-based output feedback MPC. A finite-dimensional state observer is included in the sensors to generate estimates of the slow states of the infinite-dimensional systems which are broadcast over the network to update the states of the reduced-order model used in the MPC controller at each sampling time. Precise conditions that guarantee closed-loop stability under plant-model mismatch are derived and used to devise an event-triggered sensor-controller communication strategy that minimizes network utilization without jeopardizing closed-loop stability. The key idea is to monitor the model estimation error at each sampling time and suspend communication when the prescribed stability bounds obtained based on a forecast of the future evolution of the Lyapunov function are satisfied. At times when the model estimation error fails to satisfy the projected bound on the evolution of the Lyapunov function, the sensors are prompted to proactively transmit the observer-generated state estimates to update the model states and avert instability. Finally, the design and implementation of the proposed event-triggered MPC are illustrated using a diffusion-reaction process example.