摘要
本文探讨了一阶导数、Savitzky-Golay平滑、小波变换(WT)和正交信号校正(OSC)预处理方法对混胺组分近红外光谱的预处理效果。采用预测集标准偏差(SEP)对各种预处理方法的最佳参数进行选择。比较了各种预处理方法的去噪能力,对预处理方法组合后的效果进行了评价。研究表明,WT和OSC能有效去除噪声,提高分析模型的精度,WT和OSC结合后的预处理效果更好。
In the paper,several methods including first derivative,Savitzky-Golay smooth,wavelet transform（WT）and orthogonal signal correction（OSC）were used to pretreat IR spectral data and to establish the calibration model by partial least-squares（PLS）for estimating the contents of main constituents in mixed amine.The optimum pretreatment parameters were selected through the standard error of prediction（SEP）to form optimal models.The results showed that WT and OSC were the effective methods and the combination of WT and OSC（WT-OSC）is the best choice for eliminating the noise and enhamcing the precision.
出处
《分析科学学报》
CAS
CSCD
北大核心
2010年第5期551-554,共4页
Journal of Analytical Science
关键词
近红外
预处理
混胺
小波变换
正交信号校正
Near infrared spectra
Pretreatment
Wavelet transform
Orthogonal signal correction
