【作者】 王利循；

【导师】 郭杭；

【作者基本信息】 南昌大学 ， 水利水电工程， 2010， 硕士

【摘要】 惯性导航系统本质上是非线性的,而采用低精度惯性器件将导致组合导航系统的严重非线性。常用的线性误差模型都是在假设姿态误差角是小量的情况下得到的,由于忽略了高阶项,所以当存在较大的导航误差时,线性化误差方程就不能精确地描述系统的非线性特性。而直接采用非线性模型研究IMU/GPS组合导航问题,不仅能更准确地描述系统特性,而且不需要很多约束条件。本文系统地研究了UKF非线性模型的处理算法,并基于此成功的应用于IMU/GPS组合导航系统的数据融合处理中。现把本文完成的主要内容和创新点概括如下：1、概括介绍了Kalman滤波的发展过程,对国内外Kalman滤波研究进展进行简要讨论。2、分析讨论了Kalman滤波原理和UKF非线性递推算法。并针对系统状态噪声和观测噪声为加性时,研究了UKF的改良算法。3、通过实测计算,指出了在非线性较强的IMU/GPS组合导航系统的数据处理过程中,选择UKF方法比EKF(IEKF)更具有优越性。4、针对姿态误差角对导航解的影响进行了实测数据计算分析比较。结果表明观测量中没有使用姿态误差角值情况下一般会降低了模型信息使用效率,并且在Kalman滤波计算时姿态角的结果易发散；但在观测量中加入从GPS信息中提取的姿态误差角信息,从而更好地抑制模型误差的影响,能使计算结果收敛更快,相应的精度更高。5、设计了IMU/GPS组合导航系统的无迹Kalman滤波器,利用实测数据进行了验算。更多还原

【Abstract】 Inertial navigation system is nonlinear in nature, while the low accuracy inertial navigation system will lead to a serious non-linear. Linear error models used are designed commonly in the assumption of a small attitude error angle, and because of neglecting the high order items, when there is a big navigation error, the linear error models can not accurately describe system nonlinear characteristics. Nonlinear model for direct IMU/GPS integrated navigation issues, not only can more accurately describe the system characteristics, but also not require a lot of constraints. This paper systematically studied the nonlinear model for data processing with UKF algorithms, and successfully applied on the IMU/GPS integrated navigation system data fusion processing.The main works and contributions are summarized as follows:1. The paper introduces briefly the development of Kalman filtering, and it’s current research situation both in home and abroad.2. Discussed the principle of Kalman filter and nonlinear UKF recursive algorithm. And for the system state noise and observation noise is additive, improved UKF algorithms proposed.3. According to calculation of the actually measured data, indicated that for a strong nonlinear IMU/GPS integrated navigation system data processing, selecting UKF method has more advantages than EKF (IEKF).4. Attitude error angles’impact on the navigation solution calculation and analysis of measured data is compared and analysed. The results shown that attitude error angle values can generally increased information usage efficiency of the model, and Kalman filter results calculated diverge easily without the attitude angle; but in the case of adding measurement information from the GPS attitude error angle can limit the model error much better, make the results converged faster, and improve the accuracy of the calculation.5. Designing the IMU/GPS integrated navigation system unscented Kalman filter, and checking the calculation results with the measured data.更多还原