应用近红外光谱技术对植物油酸值预测的研究  

Predicted on content of vegetable oil esters by near infrared spectroscopy

作　　者：陈志[1] 涂斌[1] 彭博[1] 郑晓[1] 何东平[2] 

机构地区：[1]武汉轻工大学机械工程学院,武汉430023 [2]武汉轻工大学食品工程学院,武汉430023

出　　处：《食品科技》2015年第12期261-264,共4页Food Science and Technology

基　　金：国家"十一五"科技支撑计划项目(2009BADB9B08);武汉市科技攻关计划项目(2013010501010147);武汉工业学院食品营养与安全重大项目培育专项(2011Z06);武汉轻工大学研究生创新基金项目(2014CX005)

摘　　要：利用近红外光谱技术结合支持向量机对植物油脂酸值含量进行回归预测。收集大豆油、花生油等油样共37份,应用激光近红外光谱仪对油样进行光谱采集,采用标准正态变量变化、多元散射校正和正交信号校正3种不同方法进行预处理。运用网格搜索法进行参数寻优,寻找最佳参数组合(C,g),建立支持向量机回归模型进行定量预测。研究表明,经过SNV、MSC和OSC预处理数据建立的模型的惩罚因子C均只有1,大大降低了模型出现过拟合现象的概率,提高了模型的泛化能力、稳健性和预测能力;预处理方法MSC和SNV建立的SVR模型校正集相关系数R较高,均达到99%;OSC建立的SVR模型具有最佳的预测性能,预测相关系数R达到93%以上;采用激光近红外光谱技术预测植物油脂酸值含量的方法是可靠的,为实现植物油脂酸值的快速检测提供了重要的依据。The content of vegetable oil esters was predicted by using near infrared spectroscopy and support vector machine.37 oil samples,such as soybean oil,peanut oil,were collected.Three kinds of different pretreatment methods,such as the standard normal variable,multiple scatter correction and orthogonal signal correction,were adopted to get the spectral sampling of the oil samples by laser near infrared spectroscopy.The grid search method was used to optimize the parameters,and to find the best parameters combination,so that we could establish the support vector machine regression model for quantitative analysis.Studies have shown that a model was established by SNV,MSC and OSC preprocessing data,and the penalty factor C of the model were only 1.It greatly reduced the fitting probability and improved the generalization ability,predictive ability and robustness of the model;the SVR model was established by pretreatment method with MSC and SNV,it has a high sets for positive correlation coefficient R,which reached 99%.OSC established the SVR model has the best performance of prediction,forecasting correlation coefficient R of 93% or more.With laser near infrared spectroscopy to predict the acid value in vegetable oils method is reliable,in order to achieve the rapid detection of plant oil acid value provides an important basis.

关 键 词：近红外光谱 油脂酸值 预测 支持向量机 

分 类 号：TS221[轻工技术与工程—粮食、油脂及植物蛋白工程]