Research On Sensor Management Algorithm For Target Detection And Tracking

Multi-Source information fusion is an integrated procedure to manage the data and information from multiple sensors,in order to meet the need of decision and estimation tasks.Considering cost,usage,switching rate of sensors and target tracking accuracy,sensor resource management technology can be involved in information fusion system to build closed-loop fusion architecture,which can be beneficial to improve the comprehensive performance of information fusion system.Sensor management based on target detection and tracking is intended to implement the adaptive allocation of sensors by establishing a management mechanism according to environment and task need.By analyzing the task needs of tracking accuracy,timeliness and robustness,two sensor management algorithms are proposed and designed.The main research work is listed as follows:Firstly,the research background and significance of the thesis is briefly described.The current progress of information fusion,sensor management technology and the relationship of information fusion and sensor management at home and abroad are reviewed.Secondly,the function,architecture and classification of sensor resource management are reviewed.The procedure of two sensor management algorithms based on covariance control and Posterior Cramér-Rao Lower Bound(PCRLB)are separately explained.The advantages and disadvantages of two algorithms are analyzed through simulation.Thirdly,on the basis of meet the target tracking accuracy,an improved sensor management algorithm based on covariance control is proposed in consideration of the timeliness and robustness of sensor management algorithm.The algorithm determines whether the sensors used at last time step can meet the accuracy of target tracking at first,by consulting the bias of the filtering covariance matrix and the desired one,using dimensional transformation and eigenvalue calculation.Set a threshold for all eigenvalues of the dimensional uniformity matrix which is transformed from the bias matrix,then determine whether the filtering covariance satisfies expectation or not,and decide whether to maintain the current sensors for choice.The performance of the proposed algorithm is tested at MATLAB and multiple radar network simulation platform based on High Level Architecture.The simulation results demonstrate the effectiveness of the algorithm.Fourthly,to solve the problem of radar allocation for stealth targets detection and tracking in radar network,a collaborative detection and tracking algorithm for stealth targets based on Conditional Posterior Cramér-Rao Lower Bound(CPCRLB)and Novel Binary Particle Swarm Optimization(NBPSO)is proposed.The tracking accuracy is measured by the CPCRLB of the tracked targets and the NBPSO is selected to search the global optimal radar allocation scheme.Then the results of particle filtering of the selected tracking radars are fused by covariance intersection algorithm.Through the simulation and analysis of the algorithm,the effectiveness is demonstrated.Finally,the main work and further research of the thesis is summarized.