多智能体协同的工业网络分布式模型预测控制研究OA

Industrial agents encompass underlying equipment,sensors,actuators,controllers,production u-nits,and scheduling systems.To address communication delays,packet losses,and limited bandwidth in indus-trial networks,an asynchronous distributed alternating direction method of multipliers(ADMM)based on neigh-bor communication is developed and integrated into a distributed model predictive control(DMPC)framework.This framework decomposes globally coupled constraints,enabling each agent to compute iterative solutions u-sing only local models and neighboring information,thereby achieving collaborative optimization under commu-nication constraints.Results demonstrate that the proposed algorithm effectively guarantees output consensus tracking and maintains closed-loop stability under non-ideal network conditions,offering a feasible solution for distributed optimal control in industrial networks.