摘要
为了解决GPS/INS松耦合导航系统中量测噪声未知所导致的滤波精度下降的问题,提出一种基于变分贝叶斯估计的卡尔曼滤波算法(VB-KF)。该算法假设量测噪声均值为0,方差服从参数未知的逆Gamma分布。通过因式分解的自由形式分布近似状态和噪声方差的联合后验分布,采用卡尔曼滤波算法估计状态,利用变分贝叶斯理论估计噪声参数,以获得系统最优的后验分布。实验结果表明,相比于传统的K F算法,该算法可实时准确估计系统状态和噪声参数,提高了滤波精度。
Considering the gap of filtering accuracy degradationcaused by the absence of measurement noise variance in GPS/INS loosely coupled navigation system,a Kalman filtering algorithm based on variational Bayesian inference(VBKF)is proposed.The algorithm assumes that the mean of measurement noise is zero,and the variance obeys the inverse Gamma distribution with unknown parameters.The joint posterior distribution of the state and noise variance is obtained by a factorized free form distribution.Kalman filter is used to estimate the state and the variational Bayesian theory is used to estimate the noise parameters to obtain the optimal posterior distribution of the system.The experimental results demonstrate that VB-KF,compared with the traditional KF algorithm,can estimate the system state and noise parameters accurately in real tim e,and improve the filtering accuracy.
作者
王艳
高嵩
马天力
陈超波
李磊
杨琼楠
Wang Yan;Gao Song;Ma Tianli;Chen Chaobo;Li Lei;Yang Qiongnan(School of Electronic and Information Engineering,Xi’an Technological University,Xi’an 710021,China)
出处
《国外电子测量技术》
2019年第11期5-10,共6页
Foreign Electronic Measurement Technology
基金
陕西省科技厅项目（2019GY-069）资助
关键词
卡尔曼滤波
逆Gamma分布
变分贝叶斯估计
组合导航
Kalman filter
inverse gamma distribution
variational Bayesian inference
integrated navigation
