论文部分内容阅读
基于红外光谱和最小二乘支持向量机建立掺杂牛奶与纯牛奶的判别模型。分别配置含有葡萄糖牛奶(0.01~0.3gL-1)和三聚氰胺牛奶(0.01~0.3gL-1)样品各36个,采集纯牛奶及掺杂牛奶样品的红外光谱。采用最小二乘支持向量机分别建立掺杂葡萄糖、掺杂三聚氰胺、两种掺杂牛奶与纯牛奶的判别模型,并利用这些模型对未知样品进行判别,其判别正确率都为95.8%。研究结果表明:与线性的偏最小二乘判别建模方法相比,最小二乘支持向量机方法具有更强的预测能力。
Distinguishing Model of Doping Milk and Pure Milk Based on Infrared Spectroscopy and Least Squares Support Vector Machine. Thirty-six glucose-free milk samples (0.01-0.3gL-1) and melamine milk samples (0.01-0.3gL-1) were respectively collected for infrared spectroscopy. The least square support vector machines were used to establish the discriminant models of doped glucose, doped melamine, two kinds of doping milk and pure milk respectively. The discriminant accuracy of the unknown samples was 95.8% using these models. The results show that the LS-SVM method has better predictive ability than the linear partial least squares discriminant modeling method.