近年来，由于相撞、侧翻和超载等原因造成校车事故不断发生。基于BD-2/GPS组合导航系统“北斗校车安全服务平台”的出现，有效的减少了校车事故的发生和中小学生人员的伤亡。本平台能够实现对校车的定位和运行状况的实时监控管理，但定位数据易受到外界环境的影响，使得定位精度变差，无法实现校车的准确定位。为此本文针对北斗校车安全服务系统对提高BD-2/GPS组合导航系统定位精度的方法进行研究，论文的主要内容包括： 本文首先给出BD-2和GPS卫星位置和速度的计算方法。然后，分别介绍BD-2和GPS的时间系统，并通过分析比较，给出了二者的转换公式。同时，对于BD-2坐标系CGS2000与GPS坐标系WGS-84的转换给出了基本方法。最后，建立基于伪距和伪距率的BD-2/GPS组合导航的数学模型，并且对组合导航的定位性能进行评估。 针对基于伪距和伪距率组合非线性观测模型，深入研究卡尔曼滤波基本理论，分析比较常用的两种非线性滤波方法，即扩展卡尔曼滤波(EKF)和无迹卡尔曼滤波(UKF)。然后将EKF和UKF算法应用到BD-2/GPS组合导航系统中，通过仿真验证UKF算法比EKF算法具有更好的优越性。 针对组合导航观测值中存在粗差问题，研究基于抗差估计理论的抗差UKF算法，并将抗差UKF应用到BD-2/GPS组合导航数据处理中，与标准UKF滤波算法进行比较，结果显示，抗差UKF算法能够减小观测粗差对导航系统精度的影响。 针对组合导航中由观测粗差和模型误差共同造成的误差问题，提出一种基于抗差Helmert方差分量估计的自适应抗差UKF算法。根据观测方差分量与状态方差分量比值设计自适应因子来调节抗差UKF算法中的增益矩阵，利用该算法对BD-2/GPS组合导航数据进行导航计算。结果表明，自适应抗差UKF算法可以很好的消弱因观测粗差和动态模型不准确对导航精度的影响，使其更好的适应于恶劣的导航环境。

In recent years, the school bus accidents continue to occur which the mainly reason is the collision, rollover and overload. The school bus safety service platform based on the BD-2/GP-S integrated navigation system reduces the number of primary and secondary school students casualties caused by the school bus accidents. This platform can be achieved on the school bus positioning and real-time running status monitoring and management, but the positioning data is easily affected by external environment, which deteriorates the positioning accuracy, unable to accurately locate the position of the bus. Aimed at the school bus safety service system to improve the method of BD-2/GPS integrated navigation system positioning precision, the main contributions of the paper include: First of all, the paper analyzes the BD-2 and GPS positioning feasibility ans advantages. Secondly, the calculation method of BD-2 and GPS satellite position and velocity are given. Then, introduces two different time-systems of BD-2 and GPS, and through analysis and comparison, the conversion formula between the two time-systems is given. Meanwhile, we study the basic methods and mathematical modela of the conversion between CGS2000 of BD-2 and WGS-84 of GPS coorfinate. Finally, we establish a mathematical model of BD-2/GPS integrated navigation based on pseudo-range and pseudo-range rate, and integrated navigation positioning performance is evaluated. For the characteristics of non-linear observation model based on pseudo-range and pseudo-range rate, the basic theory of kalman filter is further studied, then analyzes two algorithms that are widely used in nonlinear filtering methods, which are extended Kalman filter (EKF) and unscented Kalman filter (UKF). Then EKF and UKF algorithms are applied to the BD-2/GPS integrated navigation system, simulation results show that UKF algorithm is more superiority than EKF algorithm. A robust UKF algorithm based on the robust estimation theory is researched aimed at the gross errors of integrated navigation observations. The robust UKF filter is used in the data processing of BD-2/GPS integrated navigation system, compared with the standard UKF filtering algorithm, results show that the robust UKF algorithm can reduce the effect of the observations gross error to the navigation system precision. An adaptive-robust UKF filtering based on the robust Helmert variance component estimation is proposed. This new algorithm is mainly used to solve the error problem caused by observation gross error and the model error. The gain matrix is adjusted of the robust UKF according to adaptive factor which designed by the ratio between the observation variance components and the state variance components, and then this algorithm is used for navigation calculation in BD-2/GPS integrated navigation data. The results show that the adaptive robust UKF algorithm can effectively weaken the influence of observation gross error and dynamic model precision influence factors to the precision of the navigation system, so that the integrated navigation system can be used in harsh navigation environment.