Nonlinear Filter For Maneuvering Target Tracking Algorithm

Maneuvering target tracking is a typical problem of dynamic system estimation, which is used widely in military and civilian. Kalman filter is an optimal filtering algotithm in the meaning of MMSE for the linear, Gaussian system. With more and more complex applications, application system which cannot meet the conditions of linear Gaussian, has higher request for target tracking algorithm. In the non-linear, non-Gaussian system, the series of Kalman filter (including Extened Kalman filter, Unscented Kalman filter) which tracking precision decreases or diverges, cannot satisfy application request. The particle filter that provides a solutions of approximate bayes for arbitrary nonlinear, non-gaussian system, has important theoretical and practical value. This paper focuses on single maneuvering target tracking in the basis of single modle of nonlinear filtering.Firstly, this paper introduces the basic principle and mathematical model of maneuvering targets, including sports model and measurement model, and analyzes these models' application field, advantages and disadvantages. Secondly, the particle filter theory,implementation method and steps are introduced according to the knowledge of target tracking. The shortcomings of the particle filter is the degradation phemomenon which can be solved by two methods:resampling and proper choosing of importace density function.This paper proposes a improving algotithm—mix important desity function. Simulation results show that:1)Particle filter is superior to extended Kalman filter.2)Mix important desity function of particle filter is better than single important density function.Standard particle filter that uses the state transfer distribution as the importance proposal distribution, fail if the the new measurements appear in the tail of the prior or if the likelihood is too peaked in comparison to the prior. In this paper,we use an unscented Kalman filter (UKF) to generate the importance proposal distribution,so called unscented particle filter(UPF).The UKF allows the particle filter to incorporate the latest observations into a prior updating routine. In addition, the UKF generates proposal distributions that match the true posterior more closely. The new filter that results from using a UKF for proposal distribution generation within a particle filter framework is called the Unscented Particle Filter (UPF).Then the UPF are applied to the IRST(infrared search and tracking)system in the infrared point target tracking,The infrared radiation power of the target received by IRSTS is introduced to target tracking for this bearing-only measurements system,the observer model is set up of the target.The performance of the algorithm is verified by simulating a highly maneuvering target tracking, and the experimental results show that the algorithm performance better in tracking dim target.