Research Of Particle Filter And Its Application In Target Tracking

With the development of modern science and engineering technology, the two theory systems in target tracking filter domain are formed, which are classic Wiener filter using frequency domain and modern Kalman filter using state space. Kalman has already replaced the classic Wiener filter theory for it can not meet the development of modern target tracking system. Kalman Filter and Extended Kalman Filter are the most typical filter algorithms in target tracking domain, of which the former is adaptive to linear system and the latter is used in nonlinear system. Kalman Filter is an optimal filter algorithm in the Minimum-Mean-Square-Error sense, meanwhile Extended Kalman Filter is a sub-optimal filter algorithm, which derived from the linearization of nonlinear system using Taylor expansion. While the non-linearity of the system is not extreme strong, EKF can achieve approximately optimal filter effect.Although the above two methods own pretty good filtering performance when system noise and observation noise are non-Gaussian, their filtering performance will descend or even diverge when non-Gaussian distribution occurs. Thus, people begin to pay attention to filtering algorithm under nonlinear non-Gaussian background. The main conventional maneuvering target tracking algorithm is Interactive Multi Model-Extended Kalman Filter, which we can call it IMM-EKF for short. Though this algorithm is adaptive to maneuver target tracking, it can't solve problems derived from non-Gaussian noise. Further research is required in target tracking with non-Gaussian noise. Research on nonlinear problems is very hot recent years, many methods are proposed and particle filter is the more outstanding one. In this paper, research on particle filter algorithm for ballistic target tracking is carried on under the main background of nonlinearity, non-Gaussian noise.Particle filter algorithm is main content of this paper, which can be implemented by a Bayesian recursion process though a Monte Carlo simulation method. Through analysis of the basic theoretic for particle filter algorithm and study on remaining problems of the algorithm, we discuss approximate Optimal Importance Density Function Particle Filter and Evolutive Particle Filter, analyze the advantage and disadvantage of the above two filters, and eventually propose a particle filter algorithm based on new sample method. The simulation result shows that the new particle filter algorithm can solve non-Gaussian maneuver targeting tracking problem primely.