基于马氏距离的联邦卡尔曼滤波在SINS/SRS/CNS导航中的应用  

Mahalanobis distance based Federated Kalman filter for SINS/SRS/CNS integrated navigation system

作　　者：高社生[1] 洪根元 高广乐 高兵兵[1] GAO Shesheng;HONG Genyuan;GAO Guangle;GAO Bingbing(College of automation,Northwest Polytechnic University,Xi’an 710000,China)

机构地区：[1]西北工业大学自动化学院,西安710000

出　　处：《中国惯性技术学报》2021年第2期141-146,共6页Journal of Chinese Inertial Technology

基　　金：国家自然科学基金(41904028)。

摘　　要：可靠的导航信息是实现飞行器精准控制的重要条件。为提高SINS/SRS/CNS组合导航系统的可靠性与精度,提出了一种基于马氏距离的自适应联邦卡尔曼滤波算法(MD-AFKF)。在子系统传感器异常而导致产生异常量测信息时,采用基于马氏距离的噪声估计方法适时调整子系统量测噪声统计特性,同时通过在信息融合和分配阶段引入自适应融合系数与分配系数,进一步衡量各子滤波器的滤波效果并调节其协方差阵,减少不准确的子滤波器估计对主滤波器的污染。最后通过仿真验证,相较于传统联邦卡尔曼滤波算法,基于马氏距离的自适应联邦卡尔曼滤波在传感器出现量测异常时,其速度和位置精度均提高了50%以上,提高了导航系统的精确性和稳定性。Reliable navigation information is an important condition to achieve accurate control of aircraft.In order to improve the reliability and accuracy of SINS/SRS/CNS integrated navigation system,a federated Kalman filter algorithm based on Mahalanobis distance(MD-AFKF)is proposed.In case of abnormal measurement information caused by sensor failure in subsystem,a noise estimation method based on Mahalanobis distance is proposed to adjust the statistical characteristics of measurement noise of subsystem in time.At the same time,the adaptive fusion coefficient and distribution coefficient are introduced in the information fusion and allocation stage to further measure the filtering effect of each sub filter and adjust its covariance matrix to reduce the pollution of inaccurate sub filter estimation to the main filter.Finally,the simulation results show that compared with the traditional federated Kalman filter algorithm,the adaptive federated Kalman filter based on Mahalanobis distance can improve the speed and position accuracy by more than 50%when the sensor measurement is abnormal,and improve the accuracy and stability of the navigation system.

关 键 词：联邦卡尔曼滤波 传感器故障 容错能力 组合导航 马氏距离 

分 类 号：U666.1[交通运输工程—船舶及航道工程]