目标跟踪应用中,针对迭代扩展卡尔曼滤波(IEKF)在模型失配和噪声时变情况下出现精度下降甚至发散的问题,提出了一种基于多重渐消因子的自适应IEKF算法。该算法首先通过一个基于正态分布的限定记忆新息协方差估值器来计算新息协方差估计值,并根据估计均方误差把多重渐消因子分配给各数据通道;再依照χ 2检验原理判断系统是否异常,仅在系统异常时才引入渐消因子;最后利用目标与观测站间的径向距离和方位角信息,实现了IEKF迭代次数的自适应控制。仿真结果表明:与传统IEKF相比,在系统模型失配时所提算法的位置、速度和加速度平均估计误差分别减少86.97%、33.18%和15.56%;在过程噪声时变时则分别减少60.35%、18.42%和6.02%;在量测噪声时变时则分别减少50.60%、18.78%和5.41%。结果表明,所提算法有效提高了滤波精度,鲁棒性也进一步提升。

To overcome the problems of accuracy degradation and divergence of the iterative extended Kalman filter (IEKF) when applying target tracking to model mismatches and noise time-variations, an adaptive IEKF algorithm based on multiple fading factors is proposed. First, a limited memory innovation covariance estimator based on the normal distribution is used to calculate the estimated value of innovation covariance and the multiple fading factors are distributed to each filtering channel according to the estimated covariance. Then, the filtering anomaly according to the χ 2 test principle is determined and the fading factors are introduced only when the system is abnormal. Finally, the radial distance and azimuth between the target and the observing station are used to determine the adaptive control of the IEKF iteration number. The simulation results show that compared with the traditional IEKF, when the system model is mismatched, the mean estimation error of position, velocity, and acceleration of the proposed algorithm is reduced by 86.97%, 33.18%, and 15.56%, respectively. When the process noise is time-varying, it is reduced by 60.35%, 18.42%, and 6.02%, respectively. When the measurement noise is time-varying, it is reduced by 50.60%, 18.78%, and 5.41%, respectively. Therefore, the proposed algorithm effectively improves filtering accuracy and robustness.