针对粗差定值定位问题,该文从观测值与其估值的解析关系出发,提出了一种基于均值漂移模型的粗差定值定位研究思路和方法。首先利用二次投影公式和帽子矩阵的性质,推导出了参数分量和粗差估计值的另一种估计表达式,讨论了相应估计量的统计性质。然后,根据不同情况,构造了粗差检验统计量,并给出了相应的粗差检验步骤。最后,用一组包含不同粗差的模拟算例验证了该文方法的有效性。从理论上丰富了粗差探测的方法,扩大了均值漂移模型的应用范围。
Aiming at the problem of locating and evaluating gross error,this paper proposed a method and idea that could be used for this problem based on the mean shift model from the relationship between observed data and its estimation.Firstly,a new expression of the parameter estimation and the gross error estimation was deduced by utilizing the property of the re-projection formula and hat matrix,and some statistical properties of the corresponding estimator got further discussions.Secondly,test statistics of gross error were obtained from different conditions,and the procedure of checking out gross error was also provided.Finally,an example containing various gross errors demonstrated the effectiveness of the method.Theoretically this approach enriched the gross error detection method,expanded the application range of the mean shift model.