An Improved Robust Distributed Model Predictive Control Based on Adaptive Feedback Weight

Abstract

This paper proposes an improved distributed robust model predictive control algorithm based on adaptive feedback weight for the polyhedral uncertain system with input saturation. This strategy decomposes the system into several subsystems but retains independent states, and assigns robust model predictive controllers to these subsystems of which the control goals are global optimization. The feedbacks of these controllers are formulated with the subsystems’ independent states and the global state. In the presence of the control strategy, the weight changing the proportion of these states adaptively in the feedbacks can adjust the degree of coupling among the subsystems and hysteresis, so the subsystems transition from fast response to optimal stability. In a one-step iteration, the advanced strategy can not only make the steady-state value unaffected but also respond faster than the traditional distributed model predictive control algorithm. Finally, two simulation cases serve for study in the characteristics and advantages of this algorithm.