论文部分内容阅读
目的建立不同品质食用油快速分类的中红外光谱检测方法。方法不同品质的食用油在化学组分上是存在差异的,利用中红外光谱技术全面反映和整体把握食用油的化学成分信息,并借助主成分分析(PCA)结合马氏距离法对食用油的中红外光谱图进行预处理,提取其特征信息,然后通过基于统计学习理论的支持向量机(SVM)建立相应分类模型,运用模型自动鉴别不同品质的食用油类别属性。结果实验通过从市场上随机抽取食用油样本,选取了3种不同品牌的大豆油、花生油共60个样本进行测试,分类正确率达到了100%。结论基于统计学习理论的食用油红外光谱分析方法对不同品质食用油的快速分类鉴别是有效的。
Objective To establish a rapid classification method of edible oil with different quality by mid-infrared spectroscopy. Methods There are differences in the chemical compositions of different quality edible oils. By using mid-infrared spectroscopy, the chemical compositions of edible oils can be fully reflected and grasped. By means of principal component analysis (PCA) and Mahalanobis distance method, Then the characteristic information was extracted by using the mid-infrared spectrum, and then the corresponding classification model was established by the support vector machine (SVM) based on statistical learning theory. The model was used to automatically identify the different types of cooking oil. Results In the experiment, 60 samples of 3 different brands of soybean oil and peanut oil were selected for testing by randomly selecting cooking oil samples from the market, and the classification accuracy was 100%. Conclusions The edible oil infrared spectrum analysis method based on statistical learning theory is effective for rapid classification of different quality edible oils.