基于新息异常检测的改进抗差自适应卡尔曼滤波算法

在工程应用中,量测异常及量测噪声统计特性的时变是引起标准卡尔曼滤波振荡甚至发散的主要原因。经典抗差Sage-Husa自适应滤波方案,对量测中的孤立型异常有所抵抗,并可在线估计量测噪声统计特性改善滤波效果,但当连续型异常值出现时,其滤波效果不佳。针对现有抗差Sage-Husa自适应滤波方案的不足,提出了新的改进滤波方法。在改进算法中,当检测到量测异常时采用模值更大的先验预测方差阵代替原算法中的后验估计方差阵,在估计量测噪声方差时起到放大作用,以降低异常量测权重,提高滤波精度;采用IGG方案构造了新的权函数,可在抑制异常影响的同时调节估计方差阵,以免连续异常时新息持续置零引起的滤波发散;采用标准卡尔曼滤波新息辅助异常检测的双重检测策略,避免了因量测噪声方差阵的调节引起检测阈值变化而导致的漏检率增高。仿真实验表明,与常规抗差自适应滤波算法相比,该方案可更加有效地抑制量测异常值的影响。

Improved Robust Adaptive Kalman Filter with Innovation-based Outliers Diagnosis

In engineering applications, the main reasons for the oscillation or divergence of the standard Kalman filter are that the statistical property of the measurement noise varies over time and the outliers in measurements. The traditional robust Sage-Husa adaptive filter can improve the filtering performance by estimating the measurement noise covariance and restraining the isolated outliers. However, it is not the case when continuous outliers occur. To solve these problems, an improved optimal filter is designed. When the measurement outliers are detected, the measure-ment noise covariance is estimated by a replacement of a posteriori covariance with a priori covariance whose greater modulus can increase the estimated covariance and reduce the weights of outliers. A novel weighted function allowing to simultaneously restraining the outliers and tune the a posteriori covariance is designed using IGG function, which avoids the filtering divergence caused by adjusting the innovation to zero continuously. Considering the measurement noise covariance tuning will bring about a change to the threshold and cause the rises of misdetection rate, a dual outlier detection assisted by the innovation from the standard Kalman filter is used. Compared with the common robust adaptive filter, the simulation results show that the proposed scheme can mitigate the measurement outliers effectively.