近红外光谱技术快速鉴别稻谷霉菌污染的研究

Rapid Identification of Rice Mildew Pollution by Near Infrared Spectroscopy

为了建立一种快速鉴别稻谷霉菌污染的方法,研究采用近红外光谱技术结合化学计量学方法,以150份未污染霉菌的稻谷样品和150份污染霉菌的稻谷样品为研究对象,通过剔除异常光谱和光谱预处理,采用偏最小二乘回归法建立鉴别模型。结果表明:运用基于马氏距离的主成分分析方法剔除异常光谱36个,最佳光谱预处理方式为分位数标准化处理,采用基于联合x-y距离的样本集划分法,将剩余264份样品划分成训练集和验证集。建立的鉴别模型,最佳主成分数为4,其R^(2)_(cv)值为0.9220、R^(2)_(val)值为0.9184和正确率为98.48%。将外部验证集样品的光谱,代入建立并优化好的鉴别模型中,判定正确率为100%。因此,该研究所建立的鉴别模型识别能力强,可以用于稻谷中霉菌污染的快速检测。

In order to establish a method to rapidly identification of rice mold contamination,this study,150 rice samples of uncontaminated molds and 150 rice samples of contaminated molds were studied by near infrared spectroscopy combined with stoichiometry.By eliminating abnormal spectra and preprocessing the spectra,the identification model was established by partial least squares regression.The results showed that 36 abnormal spectra were eliminated by the principal component analysis based on mahalanobis distance.The best method of spectral preprocessing was Quantile Normalization.The remained 264 samples were divided into training sets and verification sets by the sample set partitioning method based on joint X-Y distance.In the established identification model,the optimal principal component number was 4,the R^(2)_(cv) value was 0.9220,the R^(2)_(val) value was 0.9184,and the accuracy was 98.48%.When the spectra of the samples from the external verification set were brought into the established and optimized identification model,the spectra were not judged to be misidentified.Therefore,the identification model established in this study has strong recognition ability and can be used for rapid detection of mold contamination in rice.