基于高斯-艾肯特滤波的机动目标跟踪算法

针对交互式多模型(IMM)算法切换滤波模型缓慢、跟踪精度低甚至发散的问题,提出了在机动目标跟踪中使用的高斯-艾肯特滤波算法。首先,该算法确定观测模型和滤波模型集,分别构造量测方程组和滤波方程组,形成总体观测矩阵;然后,针对跟踪目标的非合作机动,提出使用卡方检验来检验滤波效果,并通过滤波控制算法实时调整滤波内存长度,使用高斯-艾肯特滤波对机动目标跟踪具有很强的灵活性,实现自适应跟踪;最后,在目标跟踪仿真中与三种改进模型集的卡尔曼滤波IMM算法进行对比验证,对两类算法进行了复杂度分析。仿真结果证明了高斯-艾肯特滤波算法的有效性,在无先验信息条件下拥有更高的跟踪精度。

Maneuvering Target Tracking Method Based on Gauss-Aikten Filter

To solve the problem that the model switching speed and accuracy of standard interacting multiple-model ( IMM) algorithm are tend to decrease in maneuvering target tracking, a maneuvering target tracking algorithm using Gauss-Aikten filer is proposed. Firstly, state transition equations, measurement equations and total observation equations of Gauss-Aikten filter are derived. And then, total observation matrix is constructed via the alterable filter model and filter memory adaptively based on measurement data in each filter cycle. Finally, Gauss-Aikten filter have good flexibility on tracking maneuvering target. Performance of the Gauss-Aikten filter is evaluated via a scenario for maneuvering target tracking. Simulation results demonstrated the effectiveness of Gauss-Aikten filter compared with other three popular IMM algorithm.