基于FTIR光谱技术的霉菌菌种判别研究  

Research on Identification of Mold Species Based on FTIR Spectroscopy

作　　者：邹正宇 刘学斌 赵鑫 ZOU Zheng-yu;LIU Xue-bin;ZHAO Xin(Huayin Weapons Test Center,Huayin 714200,China)

机构地区：[1]华阴兵器试验中心,陕西华阴714200

出　　处：《红外》2022年第11期41-48,共8页Infrared

摘　　要：为了找出一种快速、简便、准确的方法来探究霉菌试验后武器装备表面生长的霉菌种类,按照标准的试验方法进行了霉菌试验,利用傅里叶红外光谱仪对经过不同菌种腐蚀后的样本进行了测试,获取了光谱数据,并通过对数据进行初步的主成分分析确定了光谱数据的识别区域。采用最小距离匹配、光谱角匹配、光谱信息散度、光谱协方差、主成分分析、偏最小二乘判别分析(Partial Least Squares Discriminant Analysis,PLS-DA)、随机森林等分类算法建立了识别模型。研究结果表明,随机森林算法能够很好地识别霉菌种类,准确率预期在98%以上。基于合适的分类算法,傅里叶红外光谱(Fourier Transform Infrared,FTIR)技术能够实现对菌种的有效鉴别。In order to find out a fast,simple and accurate method to explore the types of mold growing on the surface of weapons and equipment after mold test,the mold test according to the standard test method is conducted in this paper.Fourier infrared spectrometer is used to test the samples corroded by different strains.Spectral data are obtained,and the identification region of spectral data is determined by principal component analysis.The minimum distance matching,spectral angle matching,spectral information divergence,spectral covariance,principal component analysis,PLS-DA and random forest are used to establish the recognition model.The results show that mold species can be identified well by the random forest algorithm,and the accuracy is expected to be more than 98%.Based on the appropriate classification algorithm,FTIR technology can realize the effective identification of bacterial species.

关 键 词：傅里叶红外光谱 偏最小二乘判别分析 光谱信息散度 随机森林 霉菌试验 菌种识别 

分 类 号：O433[机械工程—光学工程]