Raman spectrogram identification of Russian anthracite and Vietnamese anthracite
The tariff rate of imported coal in China is subject to the corresponding rules of origin,so distin-guishing Russian anthracite from Vietnamese anthracite is helpful to prevent and control the risks of false reporting,concealment and misdeclaration.The representative samples of 18 batches of Russian anthracite and 15 batches of Vietnamese anthracite were collected.The spectral data were collected by 532 nm Raman spectrometer.The original spectra of Russian anthracite were compared with that of Vietnamese anthra-cite,it was found that the peak shape of Russian anthracite was relatively broad and gentle in the first-order Raman region.In the second-order Raman region,both Russian anthracite and Vietnamese anthracite showed two characteristic peaks near wavenumbers of 2 680 cm-1and 2 950 cm-1,but the peak intensity of Russian anthracite at 2 950 cm 1 was weaker.By peak fitting the first-order of the spectrogram and extrac-ting the characteristic parameters of the spectrogram,it was found that Russian anthracite and Vietnamese anthracite were close to each other in the average value of D1 peak intensity and G peak full width at half maximum(FWHM).However,the standard deviation of Vietnamese anthracite at D1 peak intensity and G peak FWHM was larger.The average value and standard deviation of D1 peak and G peak in other charac-teristics showed significant differences,such as the peak intensity ratio,FWHM ratio and peak shape coeffi-cient.Based on peak fitting combined with Fisher stepwise discriminant method,seven discriminant features of D1 peak intensity,D1 peak skewness,D1 peak kurtosis,G peak intensity,G peak skewness,G peak kur-tosis and G peak FWHM were extracted.Meanwhile,based on original spectral principal component analy-sis combined with Fisher stepwise discrimination method,5 principal components were extracted to estab-lish the identification model of Russian anthracite and Vietnamese anthracite,respectively.The accuracy of the model was evaluated by one-off cross-validation method,and the average accuracies of the two above models were both 100％.Wherein,the peak fitting based on chemical knowledge combined with Fisher stepwise discrimination model showed better discrimination ability.
NETL
NSTL
万方数据
