基于改进变分贝叶斯滤波的SINS/DVL/LBL组合导航技术  

SINS/DVL/LBL-integrated Navigation Technology Based on Improved Variational Bayesian Filtering

作　　者：赵俊波[1] 葛锡云[1] 成月 李锦 ZHAO Jun-bo;GE Xi-yun;CHENG Yue;LI Jin(China Ship Scientific Research Center,Wuxi 214082,China)

机构地区：[1]中国船舶科学研究中心,江苏无锡214082

出　　处：《水下无人系统学报》2021年第1期54-59,64,共7页Journal of Unmanned Undersea Systems

基　　金：江苏省自然科学基金项目(BK20180171);海南省重大科技计划项目(ZDKJ2019002).

摘　　要：为解决水下航行器捷联惯性导航系统(SINS)与多普勒计程仪(DVL)、长基线(LBL)定位设备组合导航问题,提出使用集中式滤波方案,并建立了SINS/DVL/LBL组合导航模型。在组合导航过程中,考虑到使用经典的卡尔曼滤波方法会存在由量测噪声方差时变和野值干扰而导致滤波精度下降的问题,通过将变分贝叶斯滤波与IGGⅢ权函数相结合,提出了一种改进变分贝叶斯滤波方法。组合导航仿真结果表明:文中滤波方法具有较强的自适应能力和抗野值能力,其滤波精度优于经典的卡尔曼滤波方法和变分贝叶斯滤波方法。To solve the integrated navigation problem of an undersea vehicle strapdown inertial navigation system(SINS),Doppler velocity log(DVL),and long baseline(LBL)location equipment,a centralized filtering scheme is proposed in this study and a SINS/DVL/LBL-integrated navigation model is established.During integrated navigation,using the classical Kalman filtering method leads to the problem in which the filtering accuracy is reduced due to time-varying measurement noise variance and the interference of outliers.Accordingly,an improved variational Bayesian filtering method is proposed by combining a variational Bayesian filter with an IGGIII weight function.Integrated navigation simulation is conducting using MATLAB software,and results show that the proposed filtering method has strong self-adaptive and anti-outlier abilities as well as a higher filtering accuracy than the classical Kalman filtering and variational Bayesian filtering methods.

关 键 词：水下航行器 组合导航 变分贝叶斯 IGGⅢ权函数 野值 

分 类 号：TJ630[兵器科学与技术—武器系统与运用工程] TP391.9[自动化与计算机技术—计算机应用技术]