Improved Interacting Multiple Model Particle Filter Algorithm Application For Maneuvering Target Tracking

Target tracking is a method of the question how to track the target whose motion can not be accurately described, estimated and predicted. The target tracking system is widely used in all kinds of areas, such as navigation, aerospace, and security defense where it is need to know the tracking path, positioning and interception of the target. The main content of the target tracking system is to established an accurate mathematical model to describe the target motion state and to design a filtering algorithm to estimated the target’s state.Because the target tracking problem is becoming more and more complex, it becomes hard and hard to describe the state of the target and the standard of the tracking system becomes more and more strict. Nowadays, the scientists have proposed many methods to establish the model of the moving target, such as CV, CA and CT model, and so on. Where, IMM algorithm is one of the better methods. It uses some models to describe the target’s state. Always, the model uses the KF algorithm to estimate the target’s moving state. But the KF algorithm needs the matrix of the noise is White gaussian noise matrix, which can’t content the need of today’s problem to target tracking.Particle filter algorithm is a suboptimal recursive bayesian estimation algorithms, it is done by monte-carlo simulation. It in strongly nonlinear and highly mobile target of filtering performance is very high, can into the optimal estimation precision and flexible use, easy to implement and has the characteristics of parallel computing structure and strong practicability, so it has wide application in the field of target tracking.In this paper, the combination of the interactive multiple model algorithm and the particle filter algorithm of the respective advantages of mutual multiple model particle filter algorithm was studied. Modeling method by studying target and filtering estimation principle of interacting multiple model particle filter algorithm principle of the algorithm are described in detail and analysis, and then introduces the interacting multiple model combined with particle filter algorithm is applied to the strongly nonlinear conditions and past with KF algorithm and EKF algorithm combined with the traditional interacting multiple model algorithm of filtering performance comparison, because now the interacting multiple model particle filter algorithm is not very mature, among them, there are still many problems, so need to improve in some ways. In this paper, in view of interaction mo particle filtering algorithm into how to establish the probability model updating method based on particle problem were studied. By adjusting the target state equation to adjust the likelihood function of particle weight value, making sample particle state closer to the real value, so that the posterior probability estimates more accurate estimation of maneuvering target tracking results more accurate, and the simulation experiments by comparing all kinds of target tracking method, proved the superiority of IMMPF.