中文摘要：

针对空间监视跟踪环境中对于包含角变量的状态向量估计存在精度较低的缺点，利用Gauss von Mises（GVM）多变量概率密度分布，提出一种基于矩匹配的GVM参数估计方法，并在此基础上改进GVM分布的确定性采样方法，建立针对GYM分布的递推滤波算法，该算法充分考虑了流形的内蕴结构，克服了传统滤波方法假设状态向量定义于欧氏空间及采用欧氏空间中高斯分布模型的局限性。仿真结果表明，该滤波算法能有效估计状态变量的后验概率分布，对角变量的估计精度明显优于扩展卡尔曼滤波方法（EKF）。

英文摘要：

In space surveillance tracking environment, in order to improve the estimation accuracy for the state of a space object which includes an angular variable, the Gauss yon Mises （GVM） distribution defined on S × R^n is employed, a GVM parameter estimation method is proposed, the deterministic sampling algorithm for GVM distribution is improved, and finally the GVM recursive filtering algorithm is developed. The algorithm takes into consideration the intrinsic structure of the manifold, instead of adopting the traditional Gaussian distribution assumption which the state variable is defined on Rn. Results demonstrate that the proposed GVM filtering algorithm can estimate the posterior probability distribution of the state vector effectively, and more accurate results can be achieved compared to the traditional extended Kalman filter （EKF） especially for angular variable.