一种鲁棒的约束总体最小二乘无源定位算法

A Robust Constrained Total Least Squares Algorithm for Passive Location

为解决异常误差导致的机载单站无源定位不准确问题,提出了一种鲁棒的约束总体最小二乘(RCTLS)定位算法.首先建立定位模型,构建了加权的约束总体最小二乘(WCTLS)定位准则,并给出了牛顿迭代解.然后,利用广义M估计原理构建了WCTLS准则的鲁棒极值函数,将鲁棒CTLS问题转化为对等价权函数的设计问题,并根据丹麦法构建了等价权函数.理论分析表明,RCTLS算法能够有效识别异常误差,并降低异常测量数据的权值以减小其对定位结果的影响.仿真结果显示,存在异常误差时,RCTLS算法能够获得理想的定位估值,具有较强的鲁棒性.

To solve the problem of location inaccuracy in single-observer airborne passive location caused by outliers,a robust constrained total least squares（RCTLS） algorithm was proposed.First,the airborne passive location model was established,and weighted constrained total least squares（WCTLS） location criterion was obtained,which could be solved by the Newton＇s method.Then,the robust extremal function for WCTLS criterion was built according to generalized M-estimation principle,thus converting the RCTLS problem to the design of equivalent weight function.The equivalent weight function was built based on Danish method.Theoretical analyses indicate that the RCTLS algorithm can identify outliers effectively,and reduce weights of abnormal measurement data in order to reduce their influence on location result.Simulations show that when outliers exist,the RCTLS algorithm keeps an ideal estimation,which proves that the RCTLS algorithm is more robust.