RAMAN结合PLS建立粗苯中多组分的定量模型  

Quantitative Model for Multi-component in Crude Benzol by RAMAN and PLS

作　　者：何书美[1] 王志彦 阎丽珍[1] HE Shumei WANG Zhiyan YAN Lizhen(Instrumental and Analysis Center, Hebei Normal University, Hebei Shijiazhuang 050024,China Hebei Provincial Engineering Technology Research Center of Coal-based Chemicals and Materials, Hebei Tangshan 063611,China)

机构地区：[1]河北师范大学分析测试中心,河北石家庄050024 [2]河北省煤基材料与化学品工程技术研究中心,河北唐山063611

出　　处：《河北师范大学学报（自然科学版）》2017年第2期147-152,共6页Journal of Hebei Normal University：Natural Science

基　　金：河北师范大学应用开发基金(L2013K07)

摘　　要：基于粗苯原料中有效组分不能快速检测的现状,利用化学计量学和偏最小二乘法原理,进行了粗苯中苯、甲苯、邻二甲苯、间二甲苯、对二甲苯含量预测的研究.依据粗苯的拉曼光谱和气相色谱数据,采用交互验证方法,选取合适的维数、波段及图谱处理方式,建立了数学模型.校正集中苯、甲苯、邻二甲苯、间二甲苯、对二甲苯的相关系数分别为0.958,0.980,0.957,0.971,0.974,预测均方根误差分别为1.210,0.393,0.085,0.393,0.085.预测精度能满足企业要求,检测时间从30min降到了几分钟.The principle of chemical metrology and partial least-squares regression are used to study the content prediction of benzene, toluene and o-xylene,m-xylene, p-xylene in crude benzol samples, because that the effective components in crude benzol raw materials can＇t be lastly determined. Based on the data of ramanspectroscopy and gas chromatograph, the mathematical models were established by the method of cross-validation, choosing appropriate dimensions, wave bands and the way of dealing with spectra. Correla- tion coefficient of benzene, toluene and o-xylene, m-xylene, p-xylene are 0. 958, 0. 980, 0. 957,0. 971, 0. 974,and the prediction error of root mean square are 1. 210,0. 393,0. 085,0. 393,0. 085,respectively. The prediction accuracy can satisfy the requirement of enterprise, and the testing time have reduced from 30 minutes to a few minutes.

关 键 词：粗苯 拉曼光谱 气相色谱 偏最小二乘法 校正模型 

分 类 号：O625[理学—有机化学] O657.37[理学—化学]