基于近红外光谱结合机器学习算法检测食用明胶品种溯源的研究  

Determination of edible gelatin origins based on near-infrared spectroscopy coupled with machine learning methods

作　　者：张浩[1,2] 刘振 王玲 胡建东[1,2] ZHANG Hao;LIU Zhen;WANG Ling;HU Jiandong(College of Mechanical and Electrical Engineering,Henan Agricultural University, Zhengzhou 450002,China;Henan International Joint Laboratory of Laser Technology in Agricultural Sciences,Zhengzhou 450002,China)

机构地区：[1]河南农业大学机电工程学院,河南郑州450002 [2]河南省农业激光技术国际联合实验室,河南郑州450002

出　　处：《河南农业大学学报》2021年第3期460-467,共8页Journal of Henan Agricultural University

基　　金：国家自然科学基金项目(32071890);河南省自然科学基金项目(202300410197);中国博士后科学基金面上项目(2017M612399);河南省高等学校青年骨干教师项目(2020GGJS046);河南农业大学科技创新基金项目(KJCX2018A09)。

摘　　要：采用近红外光谱结合机器学习方法,对5种不同来源的食用明胶进行鉴别。利用Savitzky-Golay平滑去噪、多元散射校正和最大最小归一化等方法对原始光谱数据进行预处理。将预处理的光谱数据划分为训练集和验证集,分别采用支持向量机(support vector machine,SVM)、随机森林(random forest,RF)和反向传播神经网络(back propagation neural network,BPNN)建立识别模型。结果表明,SVM模型、RF模型和BPNN模型的总体准确率均高达97%以上,其中BPNN模型的准确率为100%,明显优于其他2种模型,能够实现5种不同来源食用明胶的完全识别,而且其运算速度最短,更适用于明胶品种的溯源。Near-infrared spectroscopy(NIRS)combined with machine learning methods was used to identify five edible gelatins of different origins.The methods of Savitzky-Golay smoothing,multivariate scattering correction,and maximum and minimum normalization were used to preprocess the original spectral data.The preprocessed spectral data were divided into a training set and a validation set,and then support vector machine(SVM),random forest(RF),and back propagation neural network(BPNN)were utilized to establish the classification model,respectively.The results show that the overall accuracy of the SVM model,RF model,and BPNN model are higher up to 97%,where the accuracy of the BPNN model is 100%,which is significantly better than the other two models.Based on the BPNN model,5 edible gelatins of different origins can be completely identified.Moreover,the operation speed of BPNN model is the shortest,thus BPNN is regarded as the most suitable method for the identification of gelatin origins.

关 键 词：食用明胶 品种溯源 近红外光谱 机器学习 

分 类 号：TS202[轻工技术与工程—食品科学]