Robust Attitude and Positioning Alignment Methods for SINS/DVL Integration Based on Sliding Window Improvements
Robust Attitude and Positioning Alignment Methods for SINS/DVL Integration Based on Sliding Window Improvements
复制标题
DOI：
10.1109/tie.2023.3294582
复制
发表时间：
2024-07
影响因子：
7.7
通讯作者：
Xiang Xu;Yao Li;Lihua Zhu;Yiqing Yao
中科院分区：
文献类型：
--
作者：
Xiang Xu;Yao Li;Lihua Zhu;Yiqing Yao
捷联惯导系统(SINS)和多普勒测速系统(DVL)是水下机器人的主要定位系统。初始对准过程是SINS/DVL组合的第一个工作阶段。目前，SINS/DVL组合的初始对准方法还不够完善。主要问题可以概括为两个方面。一个是DVL输出的初始速度误差和离群值，它们没有同时被抑制。二是姿态对准完成后的定位不确定性。针对这两个问题，本文提出了鲁棒姿态和定位对准方法。首先构造了一种基于滑动窗口改进的矢量观测值来消除初始速度误差。其次，建立了滑动窗改进的矢量观测视速度运动模型。基于所建立的模型，提出了一种基于稳健卡尔曼滤波的参数估计方法。利用估计的参数，重构新的观测向量。这样，就完成了鲁棒姿态对准过程。第三，针对定位不确定性，提出了实时定位对准和后处理定位对准方法。最后，为了验证该方法的性能，设计了仿真和现场测试。测试结果表明，解决了存在的两个问题。
The strapdown inertial navigation system (SINS) and Doppler velocity log (DVL) are the major positioning system for underwater vehicles. The initial alignment process is the first working stage of SINS/DVL integration. Currently, the initial alignment method for SINS/DVL integration is imperfect. The major problems can be summarized in two aspects. One is the initial velocity errors and outliers of DVL outputs, which are not suppressed simultaneously. The other is the positioning uncertainty when the attitude alignment is finished. To address these two problems, robust attitude and positioning alignment methods are proposed in this article. First, a vector observation, which is based on the sliding windows improvements, is constructed for eliminating the initial velocity errors. Second, the apparent velocity motion model for the vector observation with the sliding window improvements is constructed. Based on the constructed model, a parameter estimation method, which is established by a robust Kalman filter, is proposed. Using the estimated parameters, the new observed vectors are reconstructed. Thus, the robust attitude alignment process is finished. Third, real-time and postprocessing positioning alignment methods are proposed for addressing positioning uncertainty. At last, the simulation and field tests are designed for verifying the performance of the proposed method. The test results are shown that the existing two problems are overcome.