激光近红外结合SVM的花生油掺伪定性定量分析  

Qualitative and Quantitative Analysis of Peanut Oil Adulteration by Laser Near Infrared Spectroscopy with SVM

作　　者：曾路路[1] 涂斌[1] 尹成[1] 郑晓[1] 宋志强[1] 何东平[2] 亓培实 

机构地区：[1]武汉轻工大学机械工程学院,武汉430023 [2]武汉轻工大学食品科学与工程学院,武汉430023 [3]武汉百信环保能源科技有限公司,武汉430023

出　　处：《中国粮油学报》2016年第8期126-130,137,共6页Journal of the Chinese Cereals and Oils Association

基　　金："十一五"国家科技支撑计划(2009BADB9B08);武汉市科技攻关计划(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2014年研究生创新基金(2014cx005)

摘　　要：利用激光近红外技术结合支持向量机(support vectormachines,SVM)对花生油掺伪进行定性和定量分析。使用激光近红外光谱仪采集188个掺入餐饮废弃油、大豆油、玉米油以及菜籽油的花生油样品光谱图。结果表明,建立的SVC分类模型均能实现100%的预测准确率,但经提取波长后的模型的变量变少,由全波段的451个波长数减少为136个。建立的SVR回归模型也能准确预测花生油中掺伪油的含量,其中非全波段模型参与建模变量变少,由451个降低到66个,预测精度也更高,校正集和测试集相关系数分别达到99.88%、99.90%,均方根误差都低于6.99E-4。由此可知,特征波长提取方法不仅可以减少建模变量,提高建模效率,也能够提高模型的预测能力。结果表明,运用激光近红外结合SVM可以实现花生油掺伪油脂的定性和定量分析。Qualitative and quantitative analyses of peanut oils adulteration were performed by combining near infrared （NIR）spectroscopy with support vector machines（SVM）. With NIR spectrometer, the spectra of 188 peanut oil samples adulterated with waste oil, soybean oil, corn oil and eanola oil were collected. Experiments demonstrated that with the SVC models,the accuracy of prediction reached to 100%. Upon wavelengths extraction,the wavelength number of full spectrum was reduced from 451 to 136. With the SVR regression models, the adulterated content of peanut oil could be predicted accurately, of which, the number of modeling variables for the non - full spectrum mod- els was reduced from 451 to 66 with higher accuracy of prediction. Correlation coefficients of calibration and test sets reached to 99.88% and 99.90% ,respectively. All the root- mean- square errors were lower than 6.99E-4. On this basis, the characteristic wavelengths extraction not only reduced modeling variables and improved the modeling ef- ficiency, but improved the modeling predictive capability. It was validated by results that the combination of NIR spectroscopy with SVM could realize the qualitative and quantitative analyses of adulteration of peanut oil.

关 键 词：激光近红外 花生油 支持向量机 定性和定量 后向偏最小二乘法 

分 类 号：O657.3[理学—分析化学]