基于支持向量回归的鱼粉TVB-N值电子鼻检测  

Detection of TVB-N in Fishmeal by Electronic Nose Based on Support Vector Regression

作　　者：刘辉[1] 牛智有[1] 

机构地区：[1]华中农业大学工学院,武汉430070

出　　处：《湖北农业科学》2011年第13期2749-2752,共4页Hubei Agricultural Sciences

基　　金：国家"十一五"科技支撑计划项目(2006BAD14B03)

摘　　要：挥发性盐基氮(TVB-N)是衡量鱼粉新鲜度一个十分重要的指标,探索鱼粉TVB-N快速检测方法对鱼粉品质检测具有重要意义。利用研制的电子鼻对不同新鲜度的鱼粉样本进行电子鼻数据采集,建立了电子鼻数据和TVB-N值之间的支持向量回归模型(SVR),利用预测集进行验证,并与多元线性回归(MLR)方法进行比较。结果表明,支持向量回归模型预测精度优于MLR模型,其决定系数R2、预测标准差SEP、最大相对误差RE-max、平均相对误差RE-mean分别为0.910、4.32、8.92%、1.87%。支持向量回归和电子鼻技术检测鱼粉TVB-N含量是可行、有效的方法。Total volatile basic nitrogen（TVB-N） is a very important indicator to measure the freshness of fishmeal.Exploring rapid detection of TVB-N has great significance to fishmeal quality＇s inspection.Different freshness samples of fishmeal were detected by self-developed electronic nose.Support vector regression（SVR） between electronic nose data and TVB-N value was created and was validated by prediction set.And it was compared to multiple linear regression（MLR） method.The results showed that the prediction precision of the model based on SVR superior to MLR model.The determination coefficients R2,standard deviation of prediction SEP,maximum relative error RE-max,and average relative error RE-mean were 0.910,4.32,8.92%,and 1.87% respectively.Therefore,it is feasible and validate to estimate TVB-N of fishmeal based on support vector regression and electronic.

关 键 词：支持向量回归 电子鼻 鱼粉 挥发性盐基氮 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]