基于姿态估计的捷联惯导动机座初始对准技术研究

The inertial frame initial alignment method can deal with the severe angular disturbance, resulting in a satisfied alignment performance, which is currently the new research topic in the area of strapdown inertial navigation system (SINS) in-motion alignment. However, the inertial frame alignment method can not estimate anything other than the attitude quaternion, such as the sensor errors (bias, scale factor etc.), so it is virtually a coarse alignment method. In order to address the aforementioned drawbacks of the inertial frame alignment method, a novel inertial frame alignment model that includes the sensor errors estimate procedure is developed based on the decomposition of the attitude quaternion/matrix. Moreover, the proposed method essentially builds up an interesting link between the attitude estimation and the SINS initial alignment. Some dynamic filtering algorithms that involved in this method have also been studied, which include the unscented Kalman filter (UKF) with consideration of the norm-constraint of the attitude quaternion, the robust UKF that can handle the external disturbance and the computational efficient UKF with the consideration of the conditional linearity of the dynamic model. The aim of this subject is to develop an accurate, efficient and robust fine initial alignment method. The proposed method possesses the advantages of theoretical innovation and engineering application, whose expected findings will provide theoretical and applied support for the SINS in-motion alignment.

基于惯性系的初始对准方法可有效隔离干扰角运动的影响，能够达到较高的初始对准精度，是目前捷联惯导动机座初始对准领域一个新的研究方向。然而现有惯性系初始对准方法未对惯性器件误差进行建模，同时不能隔离线运动干扰，因此其本质上是一种粗对准方法。基于此，本项目拟扩展现有惯性系初始对准思路，从坐标系分解角度出发推导一种考虑惯性器件误差的惯性系初始对准模型；同时将姿态估计的思想引入初始对准问题中，设计相应的动态滤波算法，具体包括：考虑四元数约束的无味卡尔曼滤波算法研究；可有效抑制外界干扰的鲁棒无味卡尔曼滤波算法研究；考虑所建立初始对准模型部分线性特性的降阶无味卡尔曼滤波算法研究。最终实现捷联惯导动机座初始精对准的目标。本项目研究思路具有一定的创新性和工程实用性，其预期成果可直接应用于目前捷联惯导动机座初始对准领域。

本项目直接面向捷联惯导动机座初始对准的实际军事需求，着重针对目前广泛研究的基于坐标系分解的惯性系初始对准方法展开研究。在传统的基于姿态确定的初始对准方法的基础上，通过巧妙的坐标系定义和分解，将载体姿态运动部分考虑进状态模型以实现对陀螺仪常值漂移的建模估计，从而将捷联惯导初始对准问题转化为了一姿态估计问题。同时，对现有速度积分公式进行改造，从而可以根据加速度计的性能合理选择积分区间长度以构造当前时刻的观测量，保证了姿态估计算法的收敛速度和精度。为了满足四元数在UKF框架中的规范性约束条件，将UKF的四元数应用形式——Unscented四元数估计器（UnScented QUaternion Estimation，USQUE）算法引入到所建立的姿态估计问题中，实现了捷联惯导的整个初始对准流程。在基于姿态确定初始对准方法基本失效的低精度捷联惯导中，所研究方法同样有效。同时，将姿态估计对准的思想引入到里程计和计程仪辅助捷联惯导初始对准问题之中，着重考虑了不同观测信息的特点，给出了算法适配性方案，实测数据实验验证了所研究算法的有效性。在本项目的资助下，发表SCI国际期刊论文14篇，EI论文2篇，协助指导毕业硕士研究生3名，博士研究生2名。