A data-driven distributed control method for performance optimization of interconnected industrial processes

In this paper, distributed plug-and-play (PnP) optimization control method for interconnected industrial processes is studied. Due to the influence of information interaction between subsystems, the performance of each subsystems are affected by their neighbor subsystem. Decentralized control approaches face difficulty in achieving satisfactory performance where the impact of information interaction between subsystems. Centralized optimal control method can achieve good results, but it burden on communication and computing. Compared with centralized and decentralized controller, distributed approaches has less online computational load and more flexible implementation schemes in interconnected industrial processes. In this work, a residual-driven distributed PnP optimization control approach is further developed, in which local residuals and residuals from neighbor subsystems are used to drive PnP optimization controllers. The influence between adjacent subsystems is fully considered in this design method. Experimental results on a three tank benchmark system show the effectiveness of the proposed algorithm.