Research On Dynamic Target Surrounding Control Of Multi-agent Systems

The major feature of swarm intelligence is that simple individual intelligence and interaction can stimulate the emergence of complex swarm intelligence behaviors.As one of the representation models of swarm intelligence,multi-agent system(MAS)has become an important research branch of swarm intelligence.In recent years,with the emerging application potential of MAS surrounding control in diverse missions of unmanned systems,such as hunting,escort,surveillance,defense,and other tasks,it has received much attention from various researchers and institutions.Therein,the key problem of dynamic target surrounding control not only lies in the design of the distributed control protocol but also lies in input constraint,system heterogeneity,and discontinuous inter-agent communication.The above problems have been studied in-depth in this dissertation.The main research works are listed as follows:Firstly,considering the input saturation problem,the saturation effect on the system stability and closed-loop performance is eliminated by using the low gain feedback technique and invariance set analysis.Thus,a two-stage dynamic target surrounding control algorithm is proposed by combining algebraic graph theory and matrix analysis,which includes driving the distance between the agents and the target to a circular motion pattern,and tweaking the repulsion force among agents.Then the surrounding control with evenly distributed phases is achieved.In addition,precisely switching regulation between two stages is guaranteed by establishing the relationship between the convergence error and the switching time.Secondly,considering the nonidentical state dimension and dynamics,a dynamic target surrounding control algorithm for a heterogeneous MAS is proposed.A distributer estimator which can simultaneously retrieve the target state and target dynamics is designed based on the perturbation system theory and input-to-state stability analysis.Then,with the help of the output regulation principle,the evenly spacing surrounding control of heterogeneous agents is finally achieved through the chasing and surrounding stages,which fulfills the target tracking and spatial distribution,respectively.The proposed control algorithm works for unknown states and unstable individual dynamics.Next,considering the intermittent inter-agent communication and the agents only access partial target information,an event-triggered surrounding control algorithm for multiple targets moving in bounded space is proposed.Based on the estimation-design-control framework,an event-triggered surrounding centroid estimator is developed firstly by utilizing the incomplete dynamic target information,which is free of the target velocity information and robust to the initial state setup.Thus,the distributed average tracking of multiple targets is achieved.Then the surrounding configuration is regulated by the time-invariant/time-varying formation vectors.Finally,the stably tracking and enclosing of multiple targets moving in bounded space under intermittent communication is fulfilled.Furthermore,an event-triggered surrounding control algorithm for multiple dynamic targets is proposed.Considering the surrounding control of multiple motional targets,a surrounding centroid estimator is designed firstly by synthesizing the internal model principle and dynamic event-triggered mechanism,whose feasibility is verified by non-smooth analysis and contradiction.Moreover,a fully distributed formation centroid estimator only using local information is proposed.Then an event-triggered phase regulation mechanism is designed to achieve balanced phase distribution.By designing the polar coordinate trajectory,the proposed control algorithm can achieve the monolayer/multilayer surrounding configuration with arbitrary closed patterns.To sum up,this dissertation conducts in-depth research on the dynamic target surrounding control of MAS under the actuation constraint,heterogeneity,and discontinuous communication,and proposes corresponding control algorithms.Finally,the conclusions of the dissertation are drawn.The unsolved problems and the future directions of dynamic target surrounding control are discussed and prospected.