Influence of stoichiometric method selection on the accuracy of the near infrared analysis of tobacco moisture content
Aims:This paper aims to study the impact of different chemometric methods on the accuracy of the near-infrared analysis of tobacco moisture content.Methods:The effects of different preprocessing methods(smoothing,first-order,second-order,standard normal variable[SNV],multivariate scattering correction[MSC]and their combinations)as well as different wavelength screening methods(based on moisture bands,wavelength intervals,and wavelength points)on the performance of prediction models were compared.Results:Models that only performed SNV,MSC,MSC+first-order,MSC+SNV,SNV+first-order preprocessing on the data could achieve varying degrees of RPD improvement while other methods decreased to varying degrees.In terms of wavelength screening methods,using wavelength interval-based methods could achieve good optimization results.After variable screening,594 wavelengths were obtained,which was 27.26％of the original number of wavelengths,and could improve the RPD value by 0.133 6.Conclusions:Different econometric methods will affect the accuracy of tobacco moisture content analysis.For this data,the MSC preprocessing method and the wavelength interval screening method should be used to process the data.
国家科技期刊平台
NSTL
万方数据
