Research On Consensus Problem And Formation Control Of Multi-agent Network

As one of the branches of distributed artificial intelligence,multi-agent network has attracted many researchers' attention.Moreover,the extensive application of multi-agent network in uav,sensor network,network congestion control and other fields is another reason.However,in practical application,due to the control of agents may be affected by noise,disturbance,model uncertainty and communication uncertainty,the information interaction between agents would produce time-delay.Therefore,it is of great practical significance to study the effect of time-delay on the performance of multi-agent networks.In order to improve the convergence speed of the second-order multi-agent network,this paper studies the positive and negative effects of time-delay on the convergence performance of the multi-agent network,and designs the consensus control protocol,and applies the control protocol to formation control,which is simulated by MATLAB platform.The research contents are as follows:(1)Firstly,the negative effects of time delay on second order continuous multi-agent networks are studied.When the time-delay of the system is symmetric and the system converges,the linear matrix inequalities satisfied by the time-delay and the input gain are obtained.The influence of different input gains on the convergence speed of the system is also discussed,and the conclusion is drawn that when the input gain of each agent is equal and the time-delay satisfies linear matrix inequality,the convergence rate of the system increases as the gains increases initially and decreases afterwards,and the equilibrium velocity of convergence can be obtained from the Laplace matrix and the input gain.The simulation results prove the accuracy of the conclusion.(2)Secondly,for the problem of the positive influence of time-delay on second-order continuous multi-agent networks,two control protocols based on current state and outdated state consistency are proposed.Due to the uncertainty of communication,the current state of one protocol contains random variables,and lyapunov method is used to analyze the conditions for the convergence of the system with time-delay and random variables when the system converges,as well as the effect of the expectation of random variables on the tolerance of time-delay.For another protocol,the outdated state contains feedback information,and the condition of time-delay is obtained by frequency domain analysis when the system converges.Selecting the appropriate outdated state can not only accelerate the convergence rate of the system,but also improve the robustness of the system with time-delay.The simulation results prove the accuracy of the conclusion.(3)Finally,for the formation control problem of multi-agent network,by using the formation control method of virtual follow-leader,this paper applies two consistency protocols based on the current state and the outdated state respectively to the multi-unmanned vehicle system in two-dimensional space and multi-unmanned aerial vehicle system in three-dimensional space.Simulation results show that both systems can form target formation,and the effectiveness of the control strategy and protocol is verified.