论文部分内容阅读
本文参考文献[1]等滤波器组的模型,提出了一种跟踪机动目标的改进型Kalman滤波器。它是通过判断观测残差是否出现偏值来检测目标的机动性的;仅在检测出机动的同时,才对机动加速度指令进行阻尼式最小二乘方估计,并用此估值来修正状态预测值及误差协方差;否则将按机动加速度指令为零的状态,以单个Kalman滤波器进行工作。这样才能使其稳态滤波精度和对机动的快速响应之间得到较好的兼顾。计算机仿真结果表明,本文所介绍的滤波器精度稍优于文献[1]中复杂滤波器组的精度,而计算量仅为后者的1/3.6。
This paper references [1] and other filter bank model, a tracking maneuvering target improved Kalman filter. It detects the maneuverability of the target by judging whether the observed residuals are biased or not. Damped least squares estimation of the maneuvering acceleration commands is performed only when the maneuver is detected, and the estimate is used to correct the state predictive value And the error covariance; otherwise it will work in a single Kalman filter with the maneuvering acceleration command zero. In this way, the steady state filtering accuracy and the quick response to maneuver can be better balanced. Computer simulation results show that the accuracy of the proposed filter is slightly better than the accuracy of the complex filter bank in [1], and the computational cost is only 1 / 3.6 of that of the latter.