基于支持向量机的假酒近红外光谱识别分类研究

一般的检测假酒中甲醇的化学方法虽然结果较准确,但操作复杂、费用昂贵且对实验的环境条件要求严格.为此,提出了一种基于支持向量机对掺甲醇的假酒光谱进行识别与分类的方法.采用ASD FieldSpec 3光谱仪测量了样品溶液的反射光谱;通过对反射光谱进行平滑、导数等预处理并进行相关性分析和单变量回归分析,得出假酒中甲醇光谱不被乙醇光谱掩盖的特征峰作为特征谱带;最后用特征谱带训练分类模型并得到分类结果.结果表明:以甲醇含量小于等于3%为真酒的总体分类准确度为85%,以甲醇含量小于等于5%为真酒的总体分类准确度为97.5%;证明该方法是可行的且具有较高的分类准确度.

Identification and Classification of Near-infrared Spectrum of Adulterated Wine Based on Support Vector Machine

Although usual chemical methods have more accurate results in detecting the methanol in adulterated wine, they are complex, expensive and requiring rigorous environment condition. A novel identified and classified spectrum of adulterated wine was proposed based on the support vector machine. The spectra of samples were measured by the ASD FieldSpec 3 spectrometer; reflection spectra were pretreated and correlation analysis and univariate regression analysis were carried out, so the peaks of methanol spectra as the characteristic bands which is not over shadowed by the ethanol were obtained; the characteristic bands were used to train classification model, the result was obtained. The result shows that, the classification accuracy is 85% while the content of methanol is less than or equal to 3% as the true wine, and the classification accuracy is 97.5% while the content of methanol is less than or equal to 5% as the true wine. So, this method is available and has higher classification accuracy.