An event-triggered cooperation approach for robust distributed model predictive control

Abstract In distributed model predictive control based on distributed optimization, the controllers of each subsystem cooperatively solve an optimization problem using an iterative algorithm. In this paper an event-triggered cooperation approach is presented for constrained linear dynamically coupled subsystems with additive disturbance where the controllers only cooperate with each other if required. This reduces the overall communication effort between the controllers. The cooperation is initiated by local event-triggers, which are designed in combination with a robust distributed model predictive controller. Thereby, recursive feasibility of the control law is guaranteed and stability of a set around the origin can be shown. An illustrative example demonstrates the reduction of the communication effort.