一种利用运动补偿的改进JPDA-UKF算法

An Improved JPDA- UKF Method Based on Motion Compensation

在恒虚警条件下,针对传统的航海雷达模拟器目标跟踪采用的基于不敏卡尔曼滤波的联合概率数据互联算法(JPDA-UKF)发散、复杂度高和实时性差的问题,提出了一种利用运动补偿的笛卡尔坐标下改进的JPDA-UKF滤波方法。该算法引入相邻周期回波间运动补偿提取的目标量测可信度矩阵,限制进入跟踪门相交区域中的虚假量测数量,并将软跟踪门技术应用于滑窗逻辑法实现航迹管理。仿真结果表明,所提方法径向速度误差比传统的JPDA-UKF算法与自适应的α-β滤波算法分别降低10%和20%,目标获得稳定航迹后径向速度归一化均方根误差(RMSE)比上述两种方法分别具有约10d B和15 d B的性能优势,算法的复杂度符合真实雷达的边扫描边跟踪的实时处理。

In order to meliorate divergence,high complexity and poor real- time performance of the traditional maritime target tracking using the joint probabilistic data association with the unscented Kalman filter( JPDA- UKF) under the condition of constant false alarm rate,an improved JPDA- UKF based on motion compensation Cartesian plane is proposed. The method restricts the number of false measurements falling into the intersection area of the tracking gates using the confidential- matrix produced by motion compensation between the adjacent time- scan echo image. The tracking management adopts the popular logic method combining with the function of soft validation gates. Simulation results show that in comparison with the two algorithms developed via traditional JPDA- UKF and adaptive coefficient α- β filtering,the proposed algorithm gains an improvement of 10 percent and 20 percent radial velocity error and an improvement of 10 d B and 15 d B in velocity root mean square error( RMSE) after getting stable track management,and also the complexity of the method is in accordance with that of virtual real- time radar scanning and tracking processing.