开放多智能体系统的预定义时间无扰编队控制OA

With the rapid development of intelligent and unmanned systems,open multi-agent systems(OMAS)have garnered significant attention due to their ability to adapt to dynamic changes in system membership.Unlike traditional multi-agent systems(MAS),OMAS allow agents to randomly join or leave during operation.While this dynamic nature en-hances system flexibility,it also introduces new control challenges:newly joined agents often possess random initial states,which can disrupt the formation stability among existing agents and even degrade overall system performance.To address these issues,this paper systematically investigates the predefined-time non-disruptive formation control problem in OMAS.First,the concept of"non-disruptive formation control"is clearly defined,with its core objective being to ensure that the joining of new agents does not affect the formation state of the existing agents.Specifically,the tracking errors of the exist-ing agents must remain zero after new agents join.Building upon this,a dual-layer control framework based on information isolation is proposed.The upper layer employs a designed information isolation algorithm based on the mean subsequence reduced method to locally identify and isolate agent states that may introduce disturbances.Concretely,each agent com-pares its state deviation with its neighbors,discards extreme values,thereby effectively filtering out abnormal state influenc-es from newly joined agents and ensuring that the cooperative information among existing agents remains uncontaminated.The lower layer introduces a predefined-time control strategy.By constructing a time-varying scaling function,the forma-tion tracking errors of all agents,including newly joined ones,converge to zero within a user-defined time,independent of initial conditions.This effectively guarantees that newly joined agents complete formation control within the predefined time.To validate the effectiveness and superiority of the proposed algorithm,numerical simulations are conducted.A dy-namic scenario with agents randomly joining and leaving is simulated,and the proposed algorithm is compared with exist-ing formation control methods.The results demonstrate that our algorithm completely eliminates the disturbance to the for-mation state of existing agents when new agents join,yielding smooth tracking error curves without fluctuations.Simultane-ously,newly joined agents can converge to the desired formation position quickly and accurately within the predefined time.In contrast,existing methods cause significant oscillations in the tracking errors of existing agents upon the arrival of new agents,disrupting the formation geometry.This further highlights the superior performance of the proposed algorithm.