基于在线回溯的高精度定位定向系统关键技术研究

Given the ability to measure both the attitude and velocity of carrier passively and autonomously, the inertial navigation system (INS) has received extensive applications in massive fields. However, INS determines the attitude and position by performing successive mathematical integrations of the information measured by inertial measurement unit (IMU) with respect to time, this fact enormously undermines the long-time precision of INS. For enhancing the accuracy of INS, the most commonly applied approach is the integration of INS and other navigation systems, i.e., Global Position System (GPS). In this method, the auxiliary navigation signals are used to estimate and compensate the errors in the INS with the help of kalman filter (KF) and online backtraking real-timely. To achieve the goal of highly precisely positioning, our main research is as following: 1. Concerning the fact that both the bias and scale factor errors of IMU, which are the major error sources of INS, are time varying, the changing residual errors of IMU are regarded as the unmodeled errors which can be determined via auxiliary information, the stability and validity of this method will be verified furthermore. 2. To obtain more accurate estimations, a novel INS error model is established which avoids the influence from the traditional one. Moreover, for taking full advantage of measurement, the forward and backward recursive filter and smooth has been employed. It recursively estimates the statistic characteristics of process noise and measurement noise via measurement sequences real-timely. 3. Fault isolation and detection (FID) of the navigation system is also under our consideration. To prevent the false information from deteriorating the performance of system, the navigation system, which has the ability of fault detection and self-reconstruction, is built up through the adaptive information parameter of federal KF. Based on the above-mentioned techniques, it is for sure that both the robustness and precision of the integrated navigation system would be improved significantly.

惯性导航具有自主性、隐蔽性和可获取载体完备运动状态的优点，得到了越来越广泛的应用，然而由于其误差会随使用时间增加而显著降低，限制了其在高精度导航中的应用。针对上述问题，通过构建惯性/多源信息组合定位定向系统，利用卡尔曼滤波方法和在线回溯技术对误差参数进行在线估计和补偿，有效提高系统精度。主要研究内容包括：（1）针对惯性器件零偏和刻度因子的时变特性，开展基于未建模动态误差参数自辨识线性滤波方法和参数估计稳定性分析方法研究，提高惯性导航系统关键参数在线辨识的正确性和可靠性。（2）建立基于计算系的误差模型，同时开展在线回溯解算模式下系统噪声、观测噪声和未精确建模系统模型对双向滤波以及平滑结果影响的研究，提高参数估计的精度。（3）为提升组合导航系统的容错性，开展基于模型匹配度检测的自适应动态信息因子联邦卡尔曼滤波研究，有效阻隔错误信息污染导航系统，提高组合导航系统的定位定向精度和鲁棒性。

本项目面向我国导航系统快速高精度定位定向的迫切发展需求，以地面车辆为背景，开展基于在线回溯的高精度定位定向系统关键技术研究，克服了现有研究中回溯次数缺乏确定标准、导航量测信息利用不充分等不足，为运载体快速高精度定位定向提供理论方法和技术支撑。.主要成果如下：.1. 提出一种状态可观测度分析方法。利用Kalman滤波对应的估计误差协方差矩阵定义状态收敛度，指数加权处理后通过一次函数映射得到统一闭区间内的可观测度。设计一种基于可观测度的误差补偿策略，有效提升了误差补偿的可靠性。.2. 静基座在线回溯对准方法研究与实现。构建了基于误差协方差矩阵的最优回溯迭代判据，获得不同时长数据满足系统对准精度要求下所需的最少迭代计算次数。针对大失准角条件下非线性问题，设计基于EKF的回溯对准方法，方位对准精度由0.44度提升至0.048度；通过引入最优双位置旋转法提高水平陀螺漂移的可观测性，进一步提升了对准精度，方位对准精度由0.042度提升至0.006度。.3. 动基座在线回溯对准方法研究与实现。引入里程计辅助完成在线回溯对准，构建基于Kalman滤波的捷联惯导/里程计精对准模型，利用TFS双向滤波平滑算法对正逆向对准结果进行融合修正。TFS方法将方位姿态RMSE由1.756度减小至0.176度。在逆向对准结果的基础上进行二次正向对准，将整段工作区间内方位RMSE由0.176度进一步减小至0.056度，方位对准精度由0.401度提升0.02度。设计基于多矢量优化定姿的回溯对准方法，减小了滤波过程中高维矩阵运算对实时性的影响。通过引入速度相关补偿项，并在逆向对准结果的基础上进行二次正向对准，进一步提升了对准精度。引入补偿项后的方位姿态RMSE由0.536度减小至0.105度，二次正向对准后方位姿态RMSE降至0.038度，对准精度由0.320度提升至0.073度。.申请发明专利4项；发表SCI收录论文1篇，EI收录论文6篇；培养研究生4名；获得国家科学技术进步二等奖一项；积极参加学术会议，与国内外同行深入交流。