一种对成片连续野值不敏感的鲁棒Kalman滤波

Robust Kalman filtering insensitive to continuous outlier

针对量测噪声中可能出现成片连续野值的情况，利用Ｂａｙｅｓ定理，得到了判别成片连续野值的后验概率公式，并提出了对上述野值不敏感的鲁棒Ｋａｌｍａｎ滤波。仿真实例表明：较之一般的Ｋａｌｍａｎ滤波，文中提出的方法能够较为满意地处理量测噪声中具有成片连续野值的恶劣情形。

Owing to the fact that continuous outliers may occur in noise measurement, a posteriori probability formula used to discriminate continuous outliers, using Bayes theorem, is derived and the robust Kalman filtering insensitive to the above outliers is presented. To confirm effectiveness of the procedure mentioned above, the state estimation of free-falling body motion is demonstrated by simulation, which gives very satisfactory robustness to continuous outliers.