Nonlinear Filtering Algorithm And Its Application Research

Target tracking has wide application in military and civilian field and filtering algorithm is a core part of target tracking. In view of the linear filtering problem in the gaussian background, the Standard Kalman Filtering algorithm is the optimal filtering algorithm in the sense of mean square. But, in fact, because most of the system is nonlinear, the Standard Kalman Filtering algorithm is not applicable and nonlinear filtering algorithm become a hot issue in the research of target tracking field. This paper studies the nonlinear filtering algorithm and its application in rocket, artillery and mortar(RAM) target and maneuvering target tracking.First of all, this paper studied several kinds of typical nonlinear filtering algorithm on the basis of the standard kalman filtering algorithm, including Extended Kalman Filtering(EKF), Unscented Kalman Filtering(UKF) and Particle Filtering(PF) algorithm. The performance of the three algorithms are evaluated, from the traditional performance evaluation indicators and the credibility Index(NCI, Noncredibility Index) through simulation.The second, aiming at the nonlinear problems of measurement equation, the existing Converted Measurement Kalman Filtering(CMKF) algorithms are studied, including Unbiased CMKF, Modified Unbiased CMKF and Estimate–Conditioned CMKF algorithm. On the base of it, aiming at the existing problem of these algorithms, Prediction-Conditioned CMKF(PCCMKF) is presented, which solves the compatibility problem in the derivation of the mean and covariance of the converted measurement errors and realize the unbiasedness of the state estimation and higher tracking precision and credibility.Nonlinear filtering algorithm is studied in the application of RAM target tracking. In view of the RAM target, system state equation is established. According to the radar tracking RAM target, nonlinear characteristics of measurement equation and state equation, the performance of different processing methods are compared from tracking precision, reliability and real-time performance through the simulation.Finally the nonlinear filtering algorithm is studied in the application of the maneuvering target tracking. Typical maneuvering target tracking algorithms are studied in the first place, including adjustable white noise model algorithm, CurrentStatistical(CS) Model algorithm and Interacting Multiple Model(IMM) algorithm, then the performance of the algorithms are evaluatd and compared through the simulation. On the base of it, considering the clutter factors in the tracking environment, the uncertainty of the target type, the nonlinear state transition characteristics in the process of motor. Adding nonlinear model to model set, appling the nonlinear filtering algorithm and probabilistic data association(PDA) in the IMM algorithm, the simulation results show that IMM-PDA algorithm can effectively solve the problem of maneuvering target tracking in a complicated environment.