Research On Consensus Of Markov Switched Multi-agent Systems With Set-valued Communication

The multi-agent system is composed of a large number of computing entities that can independently calculate and perceive each other.In recent years,cooperative control of multi-agent systems has been more and more widely used in various fields such as military,industry and aerospace.Through communication,cooperation and competition,the purpose of global control is finally achieved.Consensus problem is one of the most basic problems in the study of multi-agent system control.It means that under the action of the agreement algorithm,the states of all agents reach an agreement.In the actual complex network environment,external factors such as communication blockage and information packet loss will affect the change of the commnication relationship between agents.We usually use the Markov chain to reasonably model the change process as a random loss.Although the Markov model can accurately describe the loss process of the network connection,the analysis of the Markov model is more complicated,Therefore,the dynamic behavior of the multi-agent system under the switching topology is more difficult.Set-valued systems have widely existed in industries such as network systems,biomedical and industrial systems.The node output of a set-valued system cannot be accurately measured,and the only information that can be obtained is whether it belongs to one or more sets.In the set-valued communication environment,we need to use the set-valued identification algorithm to achieve high-precision estimation of the rough set-valued data,and design a consensus control protocol based on the estimation,and update the state of the agent under this control.Scholars have successfully applied two classical set-valued identification algorithms,the dual-time-scale consensus algorithm and the recursive projection algorithm,to the study of multi-agents with a fixed topology.However,there are few research results on how to use these two set-valued identification algorithms to estimate the state of multi-agent systems in random switching topology and set-valued communication environments.Based on the above considerations,this paper studies the consensus of multi-agent systems under Markov random switching and set-valued communication environments.The main work contents are as follows:1.Aiming at the multi-agent system under Markov switched directed topology and set-valued communication environment,we design a dual time-scale consensus algorithm based on the empirical measure method.Since the inaccuracy of set-valued information and the uncertainty of topology switching bring difficulties to the design of our algorithm,compared with the original algorithm,we enlarge the hold time in the algorithm.This enables the nodes to collect accurate estimates and enables the set-value identification part to meet the estimation accuracy requirements.2.For a special Markov switching topology,namely Bernoulli switching undirected topology and multi-agent system under the condition of set-valued communication,we estimate the neighbor state based on the recursive projection algorithm,and then use the estimated value to design a consensus control algorithm,Finally,through the cooperation of estimation and control,the multi-agent system can finally reach consensus.Different from the dual-time-scale consensus algorithm,the recursive projection algorithm is an online identification algorithm that does not require waiting time,so the convergence speed is faster.This paper provides solid theoretical support for consensus control of multi-agents under Markov switching topology and set-valued communication.And this paper promotes and develops the application scenarios of two set-valued identification algorithms,the dual-time-scale consensus algorithm and the recursive projection algorithm.