摘要
用一种小波软阈值核心偏最小二乘(WSTKPLS)方法同时定量分析光谱严重重叠的对苯二酚、邻苯二酚、间苯二酚及苯酚的混合物。结合小波软阈值法和核心偏最小二乘法改进回归质量。选择小波函数、小波分解水平、阈值法类型以优化除噪效果。采用小波软阈值核心偏最小二乘法和偏最小二乘法编制了两个计算程序。计算结果表明,WSTKPLS方法是成功的且优于偏最小二乘法。
A wavelet soft thresholding kernel partial least squares (WSTKPLS) method was developed to perform simultaneous quantitative analysis of the mixture of o-dihydroxybenzene, m-dihydroxybenzene, p-dihydroxybenzene and phenol with overlapping spectra. The quality of regression was improved by combining wavelet soft thresholding with kernel partial least squares(KPLS). Wavelet function, decomposition level and type of thresholding method were selected for optimizing the efficiency of denoising. Two programs,PWSTKPLS and PKPLS were designed to perform relevant calculations. Experimental results showed WSTKPLS method was successful and better than KPLS method.
出处
《石油化工》
CAS
CSCD
北大核心
2004年第11期1083-1087,共5页
Petrochemical Technology
基金
国家自然科学基金项目(29965001)内蒙古自然科学基金项目。
关键词
小波软阈值法
核心偏最小二乘法
多组分同时分析
重叠光谱
对苯二酚
邻苯二酚
间苯二酚
苯酚
<Keyword>wavelet soft thresholding
kernel partial least squares
simultaneous multicomponent analysis
overlapping spectra
o-dihydroxybenzene
m-dihydroxybenzene
p-dihydroxybenzene
phenol
