Research On Multi-filter Maneuvering Target Tracking Based On Interactive Multi-model

Maneuvering target tracking technology has been widely used in navigation,transportation,military and other related fields.With the continuous improvement of science and technology,for modern tracking system,the target generally has the characteristics of high speed and strong maneuverability.Accurate target tracking is always a challenging problem,especially for high-speed and strong maneuvering targets,which has huge technical difficulty in both theory and practice.The existing target tracking algorithms are mostly based on the interactive multi-model algorithm with different target maneuvering modes and higher tracking accuracy.The first is to improve the tracking accuracy of the target by directly improving the filtering algorithm of the interactive multi-model.The second is the appropriate model selection and switching method to eliminate the model mismatch problem caused by the target maneuver,so that the designed model is as close as possible to the target motion.The real trajectory,which in turn improves the tracking accuracy of the target.In view of this,this thesis conducts an in-depth study on multi-filter maneuvering target tracking based on interactive multi-model.The main research contents are as follows:(1)In the state estimation problem of the discrete time system,the model mismatch problem is caused by the different coordinate systems of target motion model and measurement data.Based on Iterative Extended Kalman Filter(IEKF)and Unbiased Conversion Measurement Kalman Filter(UCMKF),an Iterative Unbiased Conversion Measurement Kalman Filter(IUCMKF)algorithm is proposed in this study.The novel algorithm can modify the state estimation according to the measured value,make it approximate to the real value adaptively,and accelerate the convergence speed of the target tracking algorithm to some extent.Experiments show that when the initial error of the target is large,the newly proposed method can obtain faster convergence speed and higher target tracking accuracy.(2)In the target tracking problem with mobility or weak maneuverability,the target state information is usually not available and lacks a priori.The target tracking algorithm based on a single model can not accurately match the actual motion state of the target,resulting in low accuracy of target tracking or even loss of target.Aiming at such problems,this study combines the mature interactive multi-model algorithm with the previously proposed IUCCMF,and proposes an iterative unbiased conversion measurement Kalman filter algorithm based on interactive multi-model.The new algorithm utilizes the complementary characteristics of different models to overcome the problems of low precision and filtering divergence of single-model filtering.Finally,the feasibility and effectiveness of the algorithm are both verified by simulation analysis on the combination of four different models: Constant Acceleration(CA),Singer,"Current" Statistical(CS)and Jerk.(3)In the target tracking problem with strong maneuverability,the target has extremely uncertain factors,such as the target obtaining a relatively large speed in a short time.The existing maneuvering target models are difficult to accurately match the real trajectory of the target to some extent due to the extremely uncertain factors of the target.In order to solve this problem,a multi-filter target tracking algorithm based on interactive multi-model is proposed in this study.After parallel filtering of multiple filters,the simple convex combination fusion algorithm is used to fuse the estimated results of multiple filters,so that more accurate and reliable results can be obtained.Experiments show that multi-filter fusion multi-model tracking method has potential advantages in tracking accuracy compared to multi-model filter tracking method with single filter.