基于指数加权平均的GNSS/SINS组合导航系统Sage-Husa自适应卡尔曼滤波算法  

Sage-Husa Adaptive Kalman Filtering Algorithm for GNSS/SINS Integrated Navigation System Based on Exponential Weighted Average

作　　者：林雪原 孙炜玮 LIN Xueyuan;SUN Weiwei(School of Artificial Intelligence,Shandong Vocational University of Foreign Affairs,788 Changjiang Road,Weihai 264500,China;Naval Aeronautical University,Yantai 264001,China)

机构地区：[1]山东外事职业大学人工智能学院,山东省威海市264500 [2]海军航空大学,山东省烟台市264001

出　　处：《大地测量与地球动力学》2024年第12期1287-1292,1320,共7页Journal of Geodesy and Geodynamics

基　　金：国家自然科学基金(62076249);山东省自然科学基金(ZR2020MF154)。

摘　　要：测量噪声异常会导致GNSS/SINS组合导航系统滤波精度下降,甚至滤波发散。为解决该问题,首先提出适用于组合导航系统的Sage-Husa自适应卡尔曼滤波方法SHAKF;然后根据滤波新息协方差的理论估计值及实际估计值构建控制因子,提出测量噪声均方差突变起始时刻及结束时刻的检测方法,构建基于指数函数变化规律的遗忘因子,进而提出基于指数加权平均的Sage-Husa自适应卡尔曼滤波方法EWASHAKF;最后将EWASHAKF应用于GNSS/SINS组合导航系统,并进行仿真实验。结果表明,相对于SHAKF,EWASHAKF能够准确地跟踪测量噪声均方差的各种变化,进而提高组合导航系统的滤波精度。Abnormal measurement noise leads to lower filtering accuracy and even divergence of GNSS/SINS integrated navigation system.Thus,firstly we propose the Sage-Husa adaptive Kalman filtering method(SHAKF)suitable for integrated navigation systems.Then,we construct a control factor according to the theoretical and actual estimates of filtered innovation covariance,propose a detection method to measure the start and end time of measurement noise RMS mutation,construct a forgetting factor based on the exponential function variation rule,and propose a Sage-Husa adaptive Kalman filtering method based on exponential weighted average(EWASHAKF).Finally,we apply EWASHAKF to GNSS/SINS integrated navigation system and carry out the simulation experiment.The results show that compared with SHAKF,EWASHAKF can accurately track the variation of measurement noise RMS,and improve the filtering accuracy of integrated navigation system.

关 键 词：Sage-Husa算法 组合导航系统 自适应卡尔曼滤波算法 控制因子 遗忘因子 

分 类 号：P228[天文地球—大地测量学与测量工程]