Self-Triggered DMPC for Dynamically Coupled Systems With Reduced Conservatism

This article presents a less conservative self-triggered distributed model predictive control (ST-DMPC) algorithm for dynamically coupled systems with disturbances. In general, the triggering frequency is highly dependent on the prediction error between the actual and predicted states. However, this error tends to be overestimated due to the conservative handling of dynamic coupling and disturbances, leading to redundant triggering in resource-limited environments. To handle this problem, this work derives a tighter prediction error by predicting the future disturbances through their variations and incorporating them into state prediction models. Thus, compared to the traditional self-triggered criteria, the designed self-triggered conditions exhibit reduced conservatism and greater capability in reducing triggering frequency correspondingly. In addition, the designed ST-DMPC algorithm can achieve a tradeoff between computational usage and system performance. Sufficient conditions ensuring both recursive feasibility and closed-loop stability are derived, which are crucial for the algorithm’s implementation. Finally, two illustrative examples are provided to verify the merits of the algorithm from both qualitative and quantitative perspectives, respectively.