| 引用本文: | 崔加瑞,张礼廉,王茂松,吴文启,杜学禹.基于改进李群卡尔曼滤波的惯性/里程计组合导航方法[J].控制理论与应用,2023,40(12):2179~2186.[点击复制] |
| CUI Jia-rui,ZHANG Li-lian,WANG Mao-song,WU Wen-qi,DU Xue-yu.An improved LG-EKF for SINS/ODO integrated navigation[J].Control Theory and Technology,2023,40(12):2179~2186.[点击复制] |
|
| 基于改进李群卡尔曼滤波的惯性/里程计组合导航方法 |
| An improved LG-EKF for SINS/ODO integrated navigation |
| 摘要点击 1062 全文点击 337 投稿时间：2023-02-11 修订日期：2023-11-27 |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| DOI编号 10.7641/CTA.2023.30054 |
| 2023,40(12):2179-2186 |
| 中文关键词 卫星拒止环境 惯性/里程计组合导航 改进李群卡尔曼滤波器 |
| 英文关键词 satellite-denial environment SINS/ODO integrated navigation improved LG-EKF |
| 基金项目 国家自然科学基金项目(62203455)资助. |
|
| 中文摘要 |
| 基于李群非线性误差状态的扩展卡尔曼滤波方法(LG-EKF-i), 在卫星拒止场景应用中, 如高精度惯性/里程
计组合导航, 由于计算过程中需要将相对地心的位置矢量和速度矢量与姿态误差耦合, 导致在舍弃高阶项时容易产
生数值计算误差累积的问题, 致使导航精度下降. 本文提出一种改进的李群卡尔曼滤波器, 将参考系从惯性系改为
世界坐标系, 将投影系从地球系改为世界坐标系, 有效减小了李群卡尔曼滤波器在一阶近似过程中产生的数值计算
误差累积. 本文使用考虑了里程计和惯导之间安装偏差角、杆臂误差和里程计标度因数误差的21维卡尔曼滤波器,
对算法进行了验证. 在小初始对准误差角实验中, 4组长航时高精度惯性/里程计组合导航结果表明, 本文提出的改
进李群卡尔曼滤波方法(LG-EKF-w)相较于传统扩展卡尔曼滤波方法(EKF)、状态变换卡尔曼滤波方法(ST-EKF)以
及改进前的李群卡尔曼滤波方法(LG-EKF-i)具有更高的数值计算精度. 大初始对准误差角条件下的开环轨迹实验
进一步验证了LG-EKF-w算法的有效性. |
| 英文摘要 |
| The nonlinear error state-based extended Kalman filter method, known as Lie group extended Kalman filter
expressed in i-frame (LG-EKF-i), is employed in satellite-denial scenarios such as unmanned platform strapdown inertial
navigation system/odometer (SINS/ODO) integrated navigation. In this context, the position and velocity vectors relative
to geocentric are coupled with attitude errors, which may result in numerical inaccuracies and reduced precision when
neglecting higher-order terms. Therefore, in this paper, an improved LG-EKF is proposed. Replace the resolving frame and
reference frame with the world frame, which could effectively reduce the accumulation of numerical calculation error. The
algorithm was verified by using 21-states Kalman filter which considered the installation angles between the odometer and
the SINS, the lever-arm and the odometer scale factor error. Results of four long-time high-precision SINS/ODO integrated
navigation experiments with small initial alignment error angles show that the proposed improved LG-EKF-w has higher
numerical calculation accuracy than traditional extended Kalman filter (EKF), state transform extended Kalman filter (STEKF) and original LG-EKF-i. The superiority of the proposed LG-EKF-w is further verified by open-loop trajectory
experiments with large initial alignment error angles. |
|
|
|
|
|