近红外光谱分析技术定性部分的实现  

Realization of qualification analysis technology of NIR spectrum.

作　　者：韦锐[1] 

机构地区：[1]总后油料研究所

出　　处：《石油仪器》2001年第4期27-29,61,共3页Petroleum Instruments

摘　　要：文章主要介绍了“便携式油料质量分析仪”项目中实现近红外光谱定性分析技术常用到的几种分析方法 ,包括近红外光谱峰位鉴别法、模式识别法、人工神经网络法。同时对“便携式油料质量分析仪”项目中采用的光谱定性软件的组成情况进行了介绍。通过试验 ,模式识别法中的主成分分析法结合马氏距离法基本满足了油品定性分析要求 ,达到了油品鉴别、自动判断选择定量模型的目的。文章力求简化近红外光谱软件分析模式中的难点问题 ,达到深入浅出的目的。In this paper, analysis methods, which are used by NIR spectrum technology in the project of 'The portable POL quality analyser', are introduced. These methods include NIR spectrum peak valve distinguishment, pattern recognition, and artificial neural network. The NIR spectrum qualification software adopted by the project is interpreted at the same time. The principal component in pattern recognition companied with Mahalanobis distance can basically meet the requirement of POL qualitative analysis and achieve the aim of estimation of POL qualification and automatical choice of quantitative model. The article also tries best to simplify the baffling problems in the NIR spectrum software analysis model.

关 键 词：近红外光谱 模式识别 主成分分析 马氏距离 人工神经网络 油品鉴别 

分 类 号：TE622.14[石油与天然气工程—油气加工工程]