机动目标跟踪的完全最小二乘法输入估计

Total least square input estimation in tracking maneuver target

针对最小二乘法对目标机动进行估计时忽略系数矩阵偏差的问题,采用完全最小二乘法,同时考虑测量数据偏差和系数矩阵偏差对目标机动估计的影响,给出了机动目标跟踪中基于完全最小二乘法的输入估计方法。根据卡尔曼滤波器的一致性条件判断目标是否发生机动,使用完全最小二乘法估计出目标所发生的机动大小,并用目标机动的估计值对跟踪算法进行了补偿。仿真结果表明,基于完全最小二乘法的输入估计方法的跟踪效果要优于基于最小二乘法的输入估计方法。

Aiming at the problem of the LS（ least square） used in traditional input estimation method only considers the bias of measurement data, an input estimation method based on TLS （total least square） is presented in this paper. TLS considers not only the bias of measurement data but also the bias of coefficient matrix. First, target maneuver is judged according to the compatibility of Kalman filter. Then, the magnitude of target maneuver is estimated by TLS and the estimation is used to compensate for target tracking method. The simulating results prove that the target tracking result of the input estimation method based on TLS is better than that of LS.