均值漂移模型粗差探测法与LEGE法的比较  

Comparison between Data Snooping and LEGE

作　　者：宋力杰[1] 杨元喜[2] 

机构地区：[1]郑州测绘学院,河南郑州450052 [2]西安测绘研究所,陕西西安710054

出　　处：《测绘学报》1999年第4期295-300,共6页Acta Geodaetica et Cartographica Sinica

基　　金：国家杰出青年科学基金;国家自然科学基金;中国科学院动力大地测量开放实验室资助

摘　　要：粗差探测是测量数据处理、测量质量控制的重要研究主题之一。本文基于均值漂移模型详细分析论证了逐个粗差探测法与多维粗差同时定位定值法（ＬＥＧＥ法）的原理、方法和粗差探测过程，并利用两个算例的不同粗差个数和数值的各类情形进行了１ ０００ 次随机探测比较。由推证及计算表明，常规的探测法与多维粗差同时定位定值法在原理上基本等价，探测结果基本相近。Data Snooping or outlier detection is a main research subject in measurement data processing and measurement quality controlling. The principle and outlier detection procedures of LEGE and the Data Snooping based on the expectation drift model are discussed and analyzed in detail. Two examples with 1 000 kinds of multi outliers and different error values are detected and compared based on the two procedures. It is shown by the derivations and computations that the Data Snooping and LEGE are nearly equivalent in principle, the detection results are also approximately equivalent.

关 键 词：均值漂移模型 粗差探测 粗差定位与定值 

分 类 号：P207[天文地球—测绘科学与技术]