基于改进容积卡尔曼滤波的认知雷达跟踪算法

A Tracking Algorithm Based on Improved Cubature Kalman Filter in Cognitive Radar

在认知雷达目标跟踪过程中,由于存在初始跟踪误差及系统量测方程的非线性等原因,导致卡尔曼滤波算法性能较差。为解决上述问题,将Gauss-Newton迭代方法与容积卡尔曼滤波算法相结合,建立迭代容积卡尔曼滤波算法。算法在迭代过程中利用最新的量测信息并更新迭代过程中产生的新息方差,降低了目标初始状态的估计误差,并且减小了线性化量测方程引入的传递误差。仿真结果表明,迭代容积卡尔曼滤波算法与传统的扩展卡尔曼滤波算法、无迹卡尔曼滤波算法、容积卡尔曼滤波算法相比,在认知雷达中的跟踪精度更高,稳定性更好,对初始误差的容错性更强。结果可为雷达目标跟踪优化提供科学依据。

In the process of cognitive radar target tracking, Kalman filter algorithm may have poor tracking per- formance for the sake of great initial tracking error and nonlinear measurement equation of the system. This paper ad- dressed iterated cubature Kalman filter algorithm, which combines the Gauss -Newton iterate method with the cuba- ture Kalman filter. It took into account the contribution of the latest measurement information and the improved inno- vation eovariance in the iteration process, the estimation error of the target＇s initial state and the propagation error in- troduced by linearized measurement equation were reduced. The simulations show that the proposed algorithm a- chieves significantly higher rate of performance improvement, compared with extended Kalman filter, unscented Kal- man filter and cubature Kalman filter,in terms of tracking precision, stability and fault tolerance.