The Research Of Multi-agent Hierarchical Pinning Flocking

In recent years, research of multi-agent coordination control has received intense attention worldwide by scholars and experts from different fields. Among them, the flocking problem of multi-agent coordination control has become a hot focus. Flocking system is composed of lots of agents acting all by their owns, which represent coordinated consistence as to both the overall behavior and the nature according to simple behavior rules and through the interaction of local information. Multi-agent flocking behavior not only can describe and reveal self-organized coordination behavior of most of the biological community in nature, but it has wide application in engineering field such as multiple unmanned aerial vehicle system, multiple robot system, wireless self-organized network, etc. This paper mainly aims at the study and design of multi-agent system hierarchical pinning flocking algorithm and multi-objective pinning flocking control algorithm and further optimization of multi-agent flocking algorithm,1. The design of the multi-agent system hierarchical pinning flocking algorithm. Common multi-agent pinning flocking algorithms are divided as global pinning and random pinning. Among them, the global pinning flocking algorithm not only result in large energy consumption but difficult to maintain the original state in case of sudden incident, while random pinning one can even appear the "break-away" phenomenon of a few agents. Through insight analysis, this paper designs a hierarchical pinning algorithm, which first, divide the original multi-agent network into several sub-network based on degree of node association algorithm, then select "the more influential node" in each sub-network as "the information intelligent individual" according to the exponential evaluation algorithm of the node as to its influential extent, finally realize hierarchical control in the virtual leader-information agent-other intelligent individuals, and during the process of control, integrate the feedback coefficients adaptive adjustment strategy and acceleration feedback to realize the optimization of agent speed tracking. Theory analysis and simulation experiments show that hierarchical pinning flocking algorithm effectively improves the connectivity between multi-agent networks and reduces the power consumption of the network as well.2. Multi-objective pinning flocking control algorithm. In practical control, the multi-agent system is always required to track more targets at the same time, obviously, though centralized coordinated control could be realized, the global pinning algorithm can’t ensure the final pinning flock to each target sharply. Consequently, a hierarchical pinning algorithm is applied in multi-target flocking problem. Based on centralized coordinated control and through using local adaptive tracking strategy, we can realize local dynamic flocking of agents, and via employing exponential evaluation algorithm of the node as to its influential extent to select "information agents" which imitate external pinning action, the local agent will be lead to track corresponding target eventually. To avoid "break-away" of some agents locating at intersection, this paper designs an attraction-and-repelling potential function algorithm which has the advantage of few adjustable parameters and better efficiency. As an application, this paper gives out an example to avoid obstacles effectively when more targets tracking are required and more obstacles are existing. Theoretical analysis and simulation experiment, and comparing with the results derived from existing multi-objective algorithms show the effectiveness of the algorithm.3. Multi-agent flocking algorithm based on the strategy of optimizing speed. Under the circumstance that virtual leaders are not so clear that flocking group is very difficult to aggregate into larger clusters, this paper brings forth a flocking model on the basis of optimized strategy of integrating speed, which makes agents based on the consistency of the direction of the field rang (partial polarity) adaptively adjust its speed and direction, realizing effective integration of information and free. then use the particle swarm algorithm to calculate the optimal communication radius, and further step, analyze effect of the system parameters (including power exponent, speed, and group density) on the flocking behavior. The experimental results show that the proposed method makes all agents reach an equal maximum speed after a brief transient evolution of flocking groups, effectively enhance the flock behavior and increase the convergence probability of the system as well.