一种基于预测值量测转换的卡尔曼滤波跟踪算法

在雷达目标跟踪中，系统量测信息通常在球坐标系下获得。为了采用经典卡尔曼滤波算法实现有效目标跟踪，通常采用量测转换方法将非线性量测信息转换到直角坐标系中。针对传统量测转换方法基于量测值计算转换误差统计特性而导致的估计结果有偏问题，提出了一种基于预测值的量测转换方法，并将其与卡尔曼滤波算法相结合，获得了一种基于预测值量测转换的卡尔曼滤波跟踪算法。仿真结果表明，与现有的基于量测转换的卡尔曼滤波算法相比，该算法能在不提高运算量的情况下有效改善目标跟踪效果，跟踪精度提升约20%。

A Kalman filter algorithm for target tracking based on predicted position based unbiased converted measurements

In radar target tracking,the measurements are obtained in the spherical coordinates.In order to use classical Kalman filter to realize effective target tracking,the measurement conversion methods are usually adopted to convert nonlinear measurements into the Cartesian coordinates.In conventional measurement conversion method,the statistical characteristics are calculated conditioned on the measurements,which results in the bias of state estimation.For this problem,a measurement conversion method based on the predicted position is proposed.It is combined with Kalman filtering algorithm and a Kalman filter algorithm based on predicted position based unbiased converted measurements(PPUCM) for target tracking is obtained.Simulation results demonstrate that compared with existing measurement conversion based Kalman filter,the proposed algorithm can achieve much higher target tracking precision of 20% without the increase of computation.