Font Size: a A A

Research Based On The Multiagent Cluster Motion Control Method

Posted on:2010-09-23Degree:MasterType:Thesis
Country:ChinaCandidate:J HeFull Text:PDF
GTID:2208360275498279Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
The flocking control is a new distributed control method imitating biological accumulation movement in nature, and it is conducive to accomplish formation control,coordination control of multi-agent system. In recent years, the flocking control theory has gained great concern in the theory of intelligent systems research at home and abroad. In this paper, we studied the flocking of the multi-agent system by three characteristics of flocking, the aggregation, collision avoidance, as well as consistency. Main content as follows:In the first place,this paper discussed the assembly problem of first-order group system in the m-dimensional space. In the case of the network graph corresponding to group systems meeting the connectivity, we used artificial potential field function to derive the control law of every agent in the group system, and applied the LaSalle invariance principle to prove that every agent will have achieved the final assembly in the role of control law designed. Then we analyzed the assembly state of a class of group system that the network graph corresponding to it is a complete graph. According to the time-invariant characteristic of flock center, By choosing an appropriate Lyapunov function we proved that every agent in such group system will moved to a flock center and can be assembled to a Spherical region (Circular area)within a limited time in the role of control law designed. We verified that theorem to be proposed is correct, and according to the concept of gathering time and assembly time proposed we analyzed assembly property of two-dimensional and three-dimensional agent through computer simulation.Secomdly, in fixed topology or dynamic topology we studied the coherence movement of the second-order group system in a two-dimensional space. In the case of the network graph corresponding to group systems meeingt the connectivity, we used Lennard-Jones model to derive the control law of every agent in the group system, and then through the LaSalle invariance principle (also need to use nonsmooth analysis in dynamic topology )we respectively proved that in the case of leader or no-leader the group system can achieved consistency and meeted the needs of flocking, which are the aggregation and collision avoidance, in the role of control law designed. We verified that theorem to be proposed is correct, and according to the concept of flock time proposed weanalyzed assembly property of two-dimensional agent through computer simulation.
Keywords/Search Tags:Multi-agent system, Flocking, Lennard-Jones model, LaSalle invariance principle, Nonsmooth analysis
PDF Full Text Request
Related items