变形观测数据处理粗差的定位与剔除

Localization and rejection of gross errors in deformation observation data processing

在变形观测数据处理中,当变形值与观测误差的大小相关不显著时,对二者进行判断,需要进行严密的精度分析和统计检验。目前对数据进行检核,实现粗差定位,其主要方法为数理统计中的统计检验:对各周期观测值分别进行平差,进行F整体检验,检验结果存在超限误差时,进行局部检验。针对变形观测的特点,本文提出F_0整体检验法、F_1局部检验法实现粗差探测,不必剔除含粗差观测值,采用从验后方差理论导出的选择权迭代法计算,结合文献一算例,得出合理的结论,减少了许多计算工作量。

In the data-processing of the deformation measurement, the deformation value is almost equal to the error. In order to distinguish them, it is necessary to analyse carefully its accuracy and to test it by the method of hypothesis. Checking the observations and detecting of gross errors, are the main methods in hypothesis test. It is neccessary to adjust with all the periodic observations and do the integral test and the partial test. Due to the condition of deformation measurement, new methods of the integral test and the partial test are put forward. The observations with gross errors must not be rejected, the chosen iteration method is applied to the computation. An example of the leveling networks is introduced. After checking, a good result is obtained and can be applied in practical work.