VALIDATION AND REJECTION CRITERIA FOR INTEGER LEAST-SQUARES ESTIMATION
VALIDATION AND REJECTION CRITERIA FOR INTEGER LEAST-SQUARES ESTIMATION
复制标题
DOI：
10.1179/sre.1996.33.260.375
复制
发表时间：
1996-04
期刊：
影响因子：
1.6
通讯作者：
Shaowei Han;C. Rizos
中科院分区：
文献类型：
--
作者：
Shaowei Han;C. Rizos
摘要在GPS相位数据处理中，经常使用残差的最小二次值与次低值之间的对比检验作为模糊度解算的最终验证和拒绝准则。基于两个自由度为n的χ2-分布统计量n−t−m(Ω为模糊度浮点解残差的二次方形值)与自由度为m的R(R+Ω为模糊度固定解残差的二次方形值)之间的独立性，描述了一种新的F分布统计量。新的验证标准定义为。从不同条件下的测试结果来看，新的验证和拒绝标准似乎比传统标准执行得更好。
Abstract The contrast test between the minimum quadratic form value of the residuals and the second lowest has often been used as the final validation and rejection criteria of ambiguity resolution in GPS phase data processing. In this paper, a new F-distribution statistic is described, based on the independence characteristic between two χ2-distribution statistics with freedom n − t − m (Ω is the quadratic form value of residuals of the ambiguity-float solution) and R with freedom m (R + Ω is the quadratic form value of residuals of the ambiguity-fixed solution). The new validation criteria is defined as. From the results of tests under different conditions, the new validation and rejection criteria appear to perform better than the conventional criteria.