Asynchronous Separable Self-triggered Model Predictive Control based on Relaxed Dynamic Programming

Abstract

This paper presents an asynchronous distributed self-triggered MPC controller design strategy for large-scale constrained linear systems. Based on the so-called relaxed dynamic programming (RDP) inequality, at a triggered time that corresponding to a subspace of the whole system, the synthesis procedure allows us to locally determine both the updated MPC control action and the next triggering time for this subsystem. The subsystems update their inputs asynchronously, while the resulting self-triggered MPC control law can still preserve stability and constraint satisfaction. The derived conditions can be adapted to robust distributed event-triggered MPC implementation with continuous monitoring of the systems, when the system model is uncertain.