A distributed model predictive control algorithm with the gap metric output feedback decoupling

Abstract

A distributed model predictive control (DMPC) algorithm with the gap metric output feedback decoupling is proposed. For large-scale systems including strongly coupled subsystems and inaccessible states, we initially address the gap metric output feedback decoupling to weakly decouple the subsystems. The gap metric is used to analyze the similarity between the original subsystems and the decoupled ones so that subsystems’ interactions are decreased while their dynamic characteristics are unchanged. With the decoupling, reducing the communication between controllers will not limit control performance. Then a new transmission strategy is developed to improve computational efficiency of DMPC. At each sampling time, controllers communicate with each other only once and solve local control laws based on the previous optimal solutions. Furthermore, the output feedback is weighted as an input compensation to reduce the influence of model mismatch. Case studies on a reactor separator process demonstrate the effectiveness of the proposed method.