可以稳定快速收敛的无源时差定位系统

Stable-fast-convergence passive TDOA location system

针对无源时差定位系统实时性要求较高、初始估计误差较大的特点,分析了现有各种滤波方法在实际工程应用中的局限性,在此基础上提出对传统的迭代卡尔曼滤波(IKF)方法进行改进。从其高斯牛顿迭代逼近最大似然估计的实质入手,首先通过修正高斯牛顿迭代的目标函数得到一个基于一步高斯牛顿迭代的滤波更新公式,减小了初始估计误差对整个系统的影响,然后根据其最大似然估计的性质推导了一个新的增益计算公式,减少了求逆运算的次数,提高了整个系统的稳定性,在获得较好定位性能的同时大大提高了系统的处理速度。实验统计结果表明,改进后的IKF方法以较小的时间开销获得了与无迹卡尔曼滤波(UKF)方法相近的定位性能,但其定位性能明显优于扩展卡尔曼滤波(EKF)方法、粒子滤波(PF)方法和传统IKF方法。

Passive TDOA location system has high real time requirement and a little larger initial error. Aiming at this peculiarity, an improvement on conventional iterated Kalman filtering （IKF） is proposed based on the analysis of the application limitations of the existing filtering methods. According to the essence of IKF, i.e., using the Gauss-Newton method to approach the maximum likelihood estimation, we derive a new updated filtering formula based on the one-step Gauss-Newton method by modifying its objective function, and the influence of the initial estimated error on the system is decreased. A new gain formula is acquired using the property of the maximum likelihood estimation; the number of the inverse operations is reduced and the stability of the whole system is effectively improved. The proposed passive TDOA location system has much higher processing speed, while achieves good location performance. Statistical results of the experiments show that the improved IKF performs as well as unscented Kalman filtering （UKF） with less time consumption, and performs much better than extended Kalman filtering （EKF）, particle filtering （PF） and conventional IKF.