On Consensus Of Networked Multi-Agent Systems With Switching Topology And Time-Varying Delays

Networked multi-agent systems,as an important part of networked control systems,find extensive applications in cooperative tasking,dynamic decision-making,distributed optimization.In Networked multi-agent systems,delay and topology switching often damage system performance,causing the divergence of distributed optimization,and even bring system instability.Delay and topology switching are two key issues of consensus problem of networked multi-agent systems.This dissertation is mainly concerned about the consensus problem of networked multi-agent systems with time-varying delays and switching topology,including:Based on the vector-form seminorm and predictive control,the consensus problem for homogeneous networked multi-agent systems with switching topology and time-varying delay is investigated.First,for the class of networked multi-agent system with matrix B that has full row rank,a novel consensus protocol based on pseudo prediction is proposed;second,for the controllable networked multi-agent systems,using prospect control approach,a novel consensus protocol based on pseudo prediction is proposed;for general LTI networked multi-agent systems,based on the analysis of relationship between state and output consensus,the designing procedures are proposed for a couple of state/output synchronization problems,respectively.In the consensus analysis,the seminorm approach is extended from scalar form to vector form for general consensus analysis purpose.Results show that only a periodically joint spanning tree is required for consensus convergence,and the unstable modes are allowed in the system.Based on the previous results of consensus for homogeneous networked multi-agent systems,the consensus problem for heterogeneous networked multi-agent systems with switching topology and time-varying delays is investigated.For the first step,the direct method for the design of consensus protocol for heterogeneous networked multi-agent systems is proposed,i.e.,first transform the heterogeneous networked multi-agent systems with B that has row rank and those that are controllable into the homogeneous form,then do the protocol design and consensus analysis via the methods for homogeneous networked multi-agent systems.For the second step,the distributed method for the design of consensus protocol for heterogeneous networked multi-agent systems is proposed,i.e.,a two-staged method that synchronizes the system parameters and states simultaneously.The vector-form seminorm method is adopted for consensus analysis and the theoretic results are illustrated both numerically and in experiment.Results show that only a periodically joint spanning tree is required for consensus convergence,and the unstable modes are allowed in the system.Based on model-free adaptive control method,the output synchronization problem for a kind of networked non-linear multi-agent systems is investigated.Under the framework of model-free adaptive control,for the leaderless case,a novel consensus protocol is proposed,the consensus analysis based SIA matrix is conducted,with conditions that guarantee exponential convergence obtained,and the theoretic results are illustrated both numerically and in experiment.For the leader-follower case,depending whether the reference input is prescribed,a novel consensus protocol is proposed with consensus convergence conditions obtained,and the theoretic results are illustrated both numerically and in experiment.Results show that only a periodically joint spanning tree is required for consensus convergence.The prediction-based approach,multi-hop network,and optimization issue in the consensus protocol design and its analysis of networked multi-agent systems are investigated,respectively.Based on variable-delay predictive approach,novel consensus protocols for networked multi-agent systems with both fixed and switching topology are proposed,respectively,with consensus analysis conducted;based on pseudo predictive approach,novel consensus protocols for both a class of singular networked multi-agent systems with fixed topology and multi-agent systems with switching topology are proposed respectively,with consensus analysis conducted.For high-order integrator multiagent systems subjected to communication delay,two new protocols are proposed based on multi-hop relaying process for cases with and without delay,respectively,and necessary and sufficient conditions for consensus are obtained.For consensus problem with optimization,the results for perturbed exponentially consensus stable networks is obtained,based on which consensus protocols with dynamic topology and optimization are proposed.Analysis of set aggression,consensus and optimality are carried out for three different kinds of topology switching.