Cooperative Control And Tracking Algorithm For Multi-agent Platform Formation

With the sustainable development of multi-agent technology,its applications in various fields are more and more common.As one of the most efficient applications,multi-agent formation has been widely studied,such as the realization of cooperative ground target perception,the detection of multi-UAV formation in the air,and the region detection of multi-ship on the sea surface.In the traditional target tracking algorithm,static radar or single mobile agent is generally used to detect the target.However,due to the unclear target state and the increasing complexity of task,the traditional methods no longer meet the task requirements.The emergence of multi-agent cooperation control technology can solve these problems.Based on the multi-agent system,this paper studies the multi-agent cooperative control and target tracking algorithm.The specific research contents are as follows:(1)Based on multi-agent formation control theory,formation to form,formation maintenance and cooperative target tracking of multi-agent system are studied.Firstly,the moving target is modeled in the clutter environment.Then the improved Kalman filter algorithm is used to estimate the target state.Finally,a formation control algorithm of multi-agent formation based on the consistency theory is proposed.(2)Based on the topological structure of multi-agent system,this thesis proposes an index to measure the formation structure performance,and analyzes the effects of observation angle,missed detection probability and observation distance on formation performance,and improves formation performance by adjusting formation parameters.(3)In the distributed condition,the fusion estimation of the target measurements obtained by each multi-agent is carried out,and the correlation between the estimators to be fused is considered.This thesis studies the optimal fusion of density functions of each agent estimate,and applies the covariance intersection algorithm to the fusion estimation of multiple agents,and obtains the estimated value after the information fusion of multiple agents.Then the multi-agent system under the condition of noise correlation target tracking problem for related research,the measurement noise autocorrelation,state noise and measurement noise are studied in clutter environment.The case of both autocorrelation and cross-correlation is emphasized.In the abovecase,the problem of removing the noise correlation and tracking the moving target is studied.