针对声传感器纯方位目标跟踪应用,提出了一种基于静态多模型的距离参数化变周期扩展卡尔曼滤波算法,将时延影响转换为模型的可变周期,并通过参数在线估计的方法,估计该可变周期。距离参数化方法将整个距离空间分成若干个区间,分区间对距离进行估计,进而搜索出目标与观测平台之间的正确距离值。仿真结果表明变周期距离参数化扩展卡尔曼滤波算法能有效解决经典扩展卡尔曼滤波算法在纯方位角目标跟踪时可能出现的滤波发散现象,并能处理声音信号的传输时间延迟问题。
For the bearing-only target tracking with the measurement from acoustic sensors, a variable cycle range-parameterized extended Kalman filtering (RP-EKF) algorithm based on static multiple models is proposed. To deal with the time delay problem of signals received by acoustic sensors, the time-delay is transformed into the variable cycle of the motion model that can be estimated by an on-line parameter estimation method. The range from the observer to the target is estimated by the RP-EKF algorithm in which the entire distance space is firstly divided into several sub regions, and then the entire range is estimated in every sub region. The simulation results indicate that the variable cycle RP-EKF algorithm can resolve the filtering instability of EKF for the bearing-only target tracking and the time-delay problem.