利用不同分辨率(1,2,4,8cm^-1)的红外光谱建立对多组分气体样本(一氧化碳、二氧化碳、一氧化氮)的定量分析模型,研究了分辨率对气体定量测量精确性的影响。研究发现:对于样本中的各气体组分,在合适的分辨率下都有其最佳的测量精确性;而对于模型的整体精确性,在高分辨率1和2cm^-1以及低分辨率8cm^-1时模型都有较好的精确性,各组分相关系数(r)均值达到0.9995 以上,校正均方根误差(RMSEC)和验证均方根误差(RMSEP)的均值分别为18.36和15.43以下,而在分辨率为4cm^-1时,模型测量精确性陡然降低,r均值降到0.98966,RMSEC均值为90.37,RMSEP均值为64.33。实验结果表明:光谱分辨率对气体定量分析有明显的影响。为提高定量分析测量的精确性,需要针对不同的待测气体和应用场合采用合适的分辨率参数。

Spectra with different resolution(1, 2, 4 and 8cm^-1)of samples with combined CO/CO₂/NO were used to build models for quantitative analysis of each component. Using these models, the influence of spectral resolution on gas quantitative analysis was studied. Research data show that for each component there is a best competent resolution for its quantitative analysis. And the quantitative analysis of all involved components has good calibration accuracy with higher resolution(1/2cm^-1)and lower resolution(8 cm^-1), giving the relatively high mean of correlation coefficient of each component(r) more than 0.9995, and the mean of root mean square error of calibration (RMSEC) and root mean square error of prediction(RMSEP) of each component below 18.36 and 15.43 respectively, but the calibration accuracy of the model with moderate resolution of 4cm^-1 dropped sharply, giving the mean of correlation coefficient of each component(r) of 0.98966, and the mean of root mean square error of calibration(RMSEC) and root mean square error of prediction(RMSEP) of each component of 90.37 and 64.33 respectively. These results demonstrate that the spectral resolution has an important effect on the calibration accuracy of gas quantitative analysis model and the successful application of FTIR. As can be seen, the accuracy of gas quantitative analysis is highly dependent on the proper selection of spectral resolution. In order to improve the accuracy of gas quantitative analysis, it is very important to select suitable spectral resolution depending on different components and various application situations.