迭代无味卡尔曼滤波器

通过对无味卡尔曼滤波器(Unscented Kalman filter,UKF)的误差进行分析,提出了迭代UKF(IUKF)算法.该基本思路是用测量更新后的状态估计去重新对状态量和观测量的一步预测,然后再次应用LMMSE估计子估计状态量的均值和协方差阵,如此多次迭代后的滤波估计输出具有更高的精度和更小的方差,故滤波器表现出更好的一致性.Monte Carlo仿真表明,IUKF主要应用于观测噪声较小的场合,其中的迭代只需进行2～3次即可.

Iterated Unscented Kalman Filter

By analyzing the error of unscented Kalman filter(UKF),the iterated UKF (IUKF)algorithm is proposed.The IUKF is used to re-estimate the one-step state and measurement predictions by virtue of the measurement updated state variable.And then the LMMSE estimator iS reused to estimate the mean and the covariance matrix of the state variable.After several iterations,the outputs of the IUKF have higher accuracy,less variance and better consistency.The Monte carlo simulation demonstrates that the IUKF is aimed at the application with the low measurement noise and the iteration number only needs 2 or 3 time.