The Research Of Consensus In Multi-agent Systems On Complex Network

In recent years,with the rapid development of information communication,computer science or distributed networks,Multi-agent systems and CPS are used to accomplish some complex tasks,such as,Unmanned aerial vehicle group,smart grid and distributed mobile sensor network.The complexity of multi-agent systems is increasingly,especially the multi-agents are also affected by the structure of complex networks(including physical networks and information interaction networks).How to effectively improve the consensus convergence of rate for multi-agent systems is a challenging research topic in multi-agent system research.In this paper,the combination of mathematical theory and computer simulation technology,such as algebraic graph theory and matrix theory,is used to study the consensus convergence of rate for multi-agent systems on several complex networks.In the research method,the Laplacian matrix is used to describe the topology of the complex network where the multi-agent is located,and then the minimum non-zero eigenvalue of the Laplacian matrix is used to measure the consensus convergence of rate for multi-agent systems.Based on the mathematical modeling of the complex network,it is found through computer simulation that different complex network structure models lead to different convergence speeds of multi-agent systems and different factors.The improved methods of consensus convergence of rate on the different complex networks are listed as follows: for the nearest-neighbor coupled network,we can reduce the number of nodes N or increase the number of coupling K;for the Waxman random graph network,we can increase the number of the nodes N or increase the network parameter ? and ?,the convergence rate is linear with the increasing ? generally,but there will be small fluctuations;for the NW small-world network,we can increase the number of nodes N or increase the probability of random edged p,we find that the convergence rate has a good linear relationship between the number of nodes N and the probability of random edges p;In summary,the research methods and research results presented in this paper have broad reference value and guiding significance for improving the consensus convergence of rate for multi-agent systems on various complex networks.