Research On Dynamics And Control Of Flocking Of Multi-agent System

Recently, flocking, as one of the researchful focus, is a significant problem in multi-agent system research field. The results of flocking have lots of important scientific research and practical significances for people who understand the formation mechanism of nature flocking and build similar artificial intelligence systems. Because of the inspirer thinking from nature flocking, people who are interested in the information interaction of agents, dynamics modeling, forms of motion and system stability of individual for the process of group formation, and launched in-depth research for it. This thesis focuses on the connectivity preserving of network topology of flocking and the main works as follows:(1)The research on existed several main algorithms of flocking control. On the basis of development history of flocking, we explore three different models, and analyze the characteristics separately. By contrasting the characters of these three models, we chose Reynolds as the study foundation of this thesis. According to research Olfati-Saber flocking control algorithm and study the lattices, ? norm,collision function and potential function, we analyze the changing of features and parameters in detail, so that laying a solid groundwork for the next study which is about flocking control under the additional conditions.(2)The research on flocking control with connectivity preserving and multiple virtual leaders, and gain the solvable sufficiency conditions of system. Aiming at the situation of multi-target dispersion behavior in multiple virtual leaders flocking, we design a kind of distributed control protocol which is based on connectivity preserving. And then, we prove that the velocity of the center of the mass of agents will exponentially converge to weighted average velocity of the virtual leaders by the information of navigation, finally realize multi-target tribal control of flocking. Simulation results verify effectiveness of the algorithm.(3)The research on nonlinear hybrid flocking control with connectivity and event-triggered, and gain the solvable sufficiency conditions of system. Aiming at flocking with event-triggered, here we propose a kind of distributed event-triggered protocol of nonlinear hybrid flocking. At the same time, we analyze the sufficiency conditions which guarantee the multi-agent systems maintain flocking and collision avoidance under event-triggered condition. And then, according to the adding delay strategy we proved that the system, under the assumptions of initial network is connected, is always preserve network connectivity. Simulation results verify effectiveness of the algorithm.