Font Size: a A A

Research On Formation Control Strategy Of Multiple Agents Based On Dual Leaders-Follower System

Posted on:2018-01-02Degree:MasterType:Thesis
Country:ChinaCandidate:Y CaoFull Text:PDF
GTID:2322330512495236Subject:Industrial engineering
Abstract/Summary:PDF Full Text Request
With the papid development of technology in information,communication and artificial intelligence fields,the support technology and control theory of the multi-agent system have been attracted more attention from many researchers.Multi-agent system is widely used in many fields,like daily life,military and so on.For example,the most subtle way of modern mine is submarine-type mine.On the basis of the existing combat equipment,the formation control of the self-propelled radar is carried out to improve the naval combat capability and improve the efficiency of the sea blockade.It is one of the key issues in the research of the mine problem.Therefore,this thesis puts forward the research on the multi-agent formation control strategy.In the research of it,a particularly important problem is that how to make the multi-agent system form and maintain a certain formation whose initial positions are random basing on a reasonable formation control strategy.This thesis focuses on the multi-agent formation control strategy.The multi-agent system is mapped to the formation graph and the constraints between the agents can be obtained according to the formation map above.Thus,the general control rules of multi-agent are designed and these constraints are controlled to make the multi-agent system achieve the expected formation.The main results are listed as follows.Firstly,the rigid graph theory is studied and the algorithm of generating the optimal rigid graph is proposed.According to the requirements of orientation and maintainability,the optimal rigid graph mapped from the multi-agent system can be transformed into the directed formation graph based on the optimal rigid graph.Then,the basis for the following design and verification of the control law can be obtained.Secondly,the control law of multi-agent system is designed by combining the speed decomposition method and the geometric representation,and the convergence of the system under the control law can be proved.The optimal control strategy can be selected by comparing the convergence processes between the two control theories when verifying the effectiveness of the control strategies.Thirdly,considering the inevitable collision problem of multi-agent system in the process of formation convergence,the control law for the virtual gravitational field and virtual repulsion field of the multi-agent system can be designed basing on the artificial potential field method.Combined it with the control law above,the general control law of the multi-agent system formation control is formed,and the effectiveness of the control law is verified through a simulation.Finally,considering the multi-collinearity in the multi-agent system and the whole system with a straight-line formation,the directed formation map based on the optimal rigid graph is not the smallest formation Strategy between the agents.In order to improve the stability of the system,thus the rigid requirements of the formation graph can be reduced.This thesis proposes the minimum directed formation graph basing on the spanning tree.According to the general control law presented in this thesis,the rationality of the minimum contingent formation plan is proved basing on the simulation.Thus,compared with the directed formation Strategy basing on the optimal rigid graph,the control law of multi-agent system basing on the formation Strategy of minimal directed formation graph can achieve the aime that it can optimize the convergence error curve and improve the convergence speed.
Keywords/Search Tags:Multi-agent system, Directed formation graph, Leader, Follower, Collision avoidance control, Control law
PDF Full Text Request
Related items