中国食品学报2017，Vol.17Issue(8)：261-267,7.DOI:10.16429/j.1009-7848.2017.08.035
牛乳清粉掺伪羊乳粉的近红外光谱法快速无损检测
Near Infrared Spectroscopy Combined with Support Vector Regression Applied for Rapid and Nondestructive Detection of Adulterate Goat Milk Powder
邹婷婷 1王莹 2宋焕禄1
作者信息
- 1. 北京食品营养与人类健康高精尖创新中心/食品添加剂与配料北京高校工程研究中心北京工商大学 北京100048
- 2. 吉林省食品检验所 长春130022
- 折叠
摘要
Abstract
To rapidly and nondestructively detect the content of milk whey powder in adulterate goat milk powder,near infrared spectroscopy combined with v-support vector regression is applied in the study on 197 adulterate goat milk powder and the results are compared with least-squares regression method.The spectrum is pretreated with smoothing,standard variable transformation and derivative.132 samples are randomly selected as the calibration set for modeling and the remaining 65 samples as the test set for evaluation of the model performance.Results show that the optimal model of calibration set is v-SVR model pretreated with standard variable transformation,Savitzky-Golay smoothing and second derivative.Cross validation error root mean square (RMSECV) is 0.586 and cross validation correlation coefficient (RCV) is up to 0.9947.For prediction set validation results,v-SVR method is better than PLS method.Correlation coefficient of v-SVR model prediction value and real value is up to 0.9958 and RMSEP is 0.526.Results indicate that NIR combined with v-SVR model can be used to detect the content of milk whey powder in adulterate goat milk powder.It is rapid,non-destructive and easy to operate,which provides a reference for practical application.关键词
近红外光谱技术/羊乳粉/v-支持向量回归/牛乳清粉/定量分析Key words
near infrared spectroscopy/goat milk powder/v-support vector regression/milk whey powder/quantitative analysis引用本文复制引用
邹婷婷,王莹,宋焕禄..牛乳清粉掺伪羊乳粉的近红外光谱法快速无损检测[J].中国食品学报,2017,17(8):261-267,7.
