小麦霉菌污染支持向量机判别模型的建立

Rapid Identification of Mold Contamination in Wheat Using Support Vector Machine Classification

为建立一种快速判别小麦霉菌污染的方法,该研究采用近红外光谱技术结合化学计量学方法,以126份小麦样品为研究对象,通过剔除异常样品、光谱降维和预处理,采用支持向量机分类(support vector machine classification,SVM)方法建立判别模型。结果表明:运用基于马氏距离的主成分分析方法剔除异常样品5个,将原始光谱数据进行降维处理得到8个主成分,能够代表原始样本的98.80%。输入变量的最佳预处理方式为标准正态变量变换,最佳核函数为linear,核函数参数C值为10,SVM判别模型的训练集判别正确率为100%,交叉验证判别正确率为98.89%。用未参与建立判别模型的外部验证集样品对SVM判别模型进行验证,结果表明:SVM判别模型对外部验证集样品的判别正确率为100%。该研究所建立的SVM判别模型可以用于小麦霉菌污染的快速检测。

The study sought to establish a method for the rapid identification of mold contamination in wheat using 126 wheat samples.Near infrared spectroscopy combined with stoichiometry was used to establish a discriminant model based on support vector machine(SVM)classification.The SVM method was based on the elimination of abnormal samples,spectral reduction,and pretreatment.Five abnormal samples were eliminated by principal component analysis based on Mahalanobis distance.Eight principal components were obtained by reducing the dimension of the original spectral data,which represented 98.80%of the original samples.The best preprocessed method of input variables was standard normal variable transformation.The best kernel function was linear with a kernel function parameter C value of 10.The accuracy rate of training set discrimination of the SVM discriminant model was 100%and the cross-verification discrimination accuracy was 98.89%.The external verification set samples were used to verify the SVM discriminant model.The discrimination accuracy of the SVM discriminant model was 100%for the external verification set samples.The SVM discriminant model established in the study could be used for the rapid detection of mold contamination in wheat.