特征变量筛选在近红外光谱测定绿茶汤中茶多酚的应用  

Application of Characteristic Variables Selection in Determination of Polyphenols Content in Green Tea Infusion by Near Infrared Spectroscopy

作　　者：吴瑞梅[1,2] 岳鹏翔 赵杰文[1] 黄星奕[1] 陈全胜[1] 

机构地区：[1]江苏大学食品与生物工程学院,镇江212013 [2]江西农业大学工学院,南昌330045 [3]大闽食品(漳州)有限公司,漳州363005

出　　处：《农业机械学报》2011年第12期154-157,163,共5页Transactions of the Chinese Society for Agricultural Machinery

基　　金：国家自然科学基金资助项目(30971685);江苏省自然科学基金资助项目(BK2009216);福建漳州市蓝火计划资助项目(Z2010016);江苏大学博士创新计划资助项目(CX10B_020X)

摘　　要：利用化学计量学方法从绿茶汤近红外光谱中提取茶多酚光谱信息,建立茶多酚近红外光谱定量分析模型。光谱采集使用5 mm光程的石英比色皿,利用联合区间偏最小二乘法(siPLS)筛选特征光谱区间,然后在筛选的光谱区间内进一步利用遗传算法(GA)优选特征变量。结果表明,siPLS筛选的特征光谱区间避开了水的强吸收峰影响,利用GA在筛选的特征光谱区间内优选出166个特征变量建立PLS模型,模型预测集均方根误差为0.685%,相对标准差为5.26%,相对分析误差为3.22,所建模型能达到精度要求,可用于实际检测。The feasibility to extract the polyphenols spectra information from near infrared spectroscopy of green tea infusion using chemometrics methods was attempted.A cylindrical quartz glass tube with an optical path of 5 mm was used.First,synergy interval PLS（siPLS） was implemented to select efficient spectral regions from SNV preprocessed spectra.Then,optimal variables were selected using genetic algorithm（GA） from these selected spectral regions by siPLS.This study showed that the selected characteristic spectral intervals were not within the range of the strong absorbance for water.When 166 variables selected by GA and eight PLS factors were included,the optimal model（siPLS-GA） was achieved with PRMSEP=0.685%,PRSD=5.26% and PRPD=3.22 in the prediction set.The results showed that the performance of siPLS-GA model could be employed to measure polyphenols content in green tea infusion by near infrared spectroscopy in practice.

关 键 词：茶多酚 近红外光谱 联合区间偏最小二乘法 遗传算法 

分 类 号：O657.33[理学—分析化学]