一种基于软卡方检测的自适应Kalman滤波方法

在传统Kalman滤波中,卡方检测方法简单地将量测划分为正常和异常两种,针对其不足之处,改进并提出了一种新的软卡方检测方法。新方法根据卡方检测结果构造连续变化的量测,利用权重系数,充分挖掘处于正常值与异常值之间的可疑量测新息,同时还将该方法推广成多维量测的多分量卡方检测形式,建立了全面完整的软卡方检测Kalman滤波量测更新方程。最后,通过惯导/卫导组合导航仿真,验证了软卡方检测Kalman滤波的优势:无需任何参数调整且具有比Sage-Husa自适应滤波更小的统计误差波动。

An adaptive Kalman filter based on soft Chi-square test

In the traditional Kalman filter, the Chi-square test method simply divides the measurement into two categories, the normal ones and the abnormal ones. In view of its shortcomings, a new soft Chi-square test method is improved and proposed. According to the Chi-square test results, the new method constructs continuously changing measurements and uses the weight coefficient to fully explore the new suspicious measurement between the normal value and the abnormal value. At the same time, the method is extended to the multi-component Chi-square test form of multi-dimensional measurement, and a comprehensive and complete update equation of soft Chi-square test Kalman filter measurement is established. Finally, the simulation of INS/GNSS integrated navigation verifies the advantages of soft Chi-square test Kalman filter, the proposed method does not need any parameter adjustment and has smaller statistical error fluctuation than classical Sage-Husa adaptive filter.