Brown Andrew S, Milton Martin J T, Cowper Chris J, Squire Gavin D, Bremser Wolfram, Branch Richard W
Analytical Science Group, National Physical Laboratory, Queens Road, Teddington, Middlesex TW11 0LW, UK.
J Chromatogr A. 2004 Jun 25;1040(2):215-25. doi: 10.1016/j.chroma.2004.04.007.
The results of gas chromatographic analysis of natural gas mixtures reveal strong correlations (Pearson correlation coefficient of >0.96) between the uncertainty of each component and variations in the ambient pressure. Although correction for ambient pressure variations can reduce this variability, normalisation of the results of each analysis using the assumption that the sum of all component amount fractions is unity provides significantly greater reductions in the uncertainty of each measured component. We show that the uncertainty in normalised components can be estimated approximately using the correlation coefficient as a measure of the correlation present in the measurements, or exactly using a full calculation of the variance/covariance (V/C) structure of the data.
天然气混合物的气相色谱分析结果表明，各组分不确定度与环境压力变化之间存在强相关性（皮尔逊相关系数>0.96）。尽管对环境压力变化进行校正可降低这种变异性，但假设所有组分量分数之和为1对每次分析结果进行归一化处理，能显著更大程度地降低各测量组分的不确定度。我们表明，归一化组分的不确定度可近似地使用相关系数作为测量中存在的相关性的量度来估计，或者通过对数据的方差/协方差（V/C）结构进行完整计算来精确估计。
