基于方差膨胀模型的多个粗差的探测

Detection of multiple gross errors based on variance inflation model

本文将粗差归入随机模型,提出了基于方差膨胀模型的粗差探测方法。首先针对测量平差实际给出了非等权独立观测条件下的单个粗差的Score检验统计量,然后提出了基于方差膨胀模型的两种定位多个粗差的方案,最后对一边角网进行了计算和分析。大量试验表明,用本文给出的定位多个粗差的方法是切实可行的,它不仅有效地发现了粗差,而且计算简便、快捷,结果比较理想。

Ascribing gross errors to random model,this paper puts forward a detection method of multiple gross errors based on variance inflation model.Taking surveying adjustment practice into account,we first give the score test statistics on single gross error under the condition of unequal weight and individual observations.After that,we present two schemes based on variance inflation model to locate more than one gross error.Finally,as an example we compute and analyze a side-angel network.Many experiments show that the method given is feasible.This resolution is simple and the result is perfect comparative.