摘要
本文提出了基于红外光谱,结合化学计量学快速分析及无损识别轮胎橡胶颗粒的方法。实验采集不同品牌样本的红外光谱数据,通过预处理和主成分分析,实现对840个原始特征到5个识别特征的高效筛选,基于综合考虑神经元个数和网络性能,以及模型准确度的目的,输入层选择PCA1、PCA2、PCA3、PCA4和PCA5为特征变量,采用递增方法确定隐含层神经元个数,构建径向基函数神经网络(RBF)轮胎橡胶的分类模型,最终对75个样本实现了准确率为92.0%的识别分类。结果表明,利用红外光谱和径向基函数模型可实现对轮胎橡胶样本光谱较为准确的识别与分类,其可为其他物证的鉴别与分析提供一定的参考。
A method based on infrared spectroscopy and chemometrics is proposed to non-destructively and quickly identify the tire rubber particles.The different brands samples' spectra were collected,and pretreatment and principal component analysis were used to reduce the dimension from 840 original features to 5 features.Based on comprehensive consideration of the number of neurons and network performance,the selected PCA1,PCA2,PCA3,PCA4 and PCA5 as characteristic variables,and the incremental method in the hidden layer was used to construct the model.Finally,the classification accuracy rate of 75 samples is 92.0%.The result showed that combining the infrared spectra and chemometrics can non-destructively,quickly and accurately identify the tire rubber particles,which has a strong versatility and high efficiency in physical evidence identification.
引文
[1]ZHANG Y F,KONG B.China Petroleum and Chemical Standards and Quality(张艳芬,孔波.中国石油和化工标准与质量),2016,36(03):3.
[2]SUN Z W,CHENG F B,QIAO T,et al.Chin J Anal Lab(孙振文,程芳彬,乔婷,等.分析试验室),2017,36(12):1402.
[3]WEI F X.Modern Instrument Analysis Technology and Application.Beijing:China Petrochemical Press(魏福祥.现代仪器分析技术及应用.北京:中国石化出版社),2015:84.
[4]OUYANG A G,ZHANG Y,TANG T Y,et al.Laser&Optoelectronics Progress(欧阳爱国,张宇,唐天义,等.激光与光电子学进展),2017,54(9):093003.
[5]CAI J H,YING S J,CHEN C,et al.Journal of Analytical Science(蔡嘉华,尹嵩杰,陈超,等.分析科学学报),2016,32(4):458.
[6]SHI L F,QU J,XING Z N.Physical Testing and Chemical Analysis(Part B:Chemical Analysis)(史令飞,瞿军,邢志娜.理化检验-化学分册),2018,54(2):200.
[7]ZHANG Y,GUO Q,WANG J Y.Advanced Engineering Sciences(章毅,郭泉,王建勇.工程科学与技术),2017,49(1):9.
[8]HE Q,HUANG B J,CHENG J R,et al.Journal of Analytical Science(何琴,黄保军,程静茹,等.分析科学学报),2016,32(5):697.
[9]WANG X D,SHAO H H.Information and Control(王旭东,邵惠鹤.信息与控制),1997,(4):32.
[10]CHEN G Q,WU Y M,LIU H J,et al.Spectroscopy and Spectral Analysis(陈国庆,吴亚敏,刘慧娟,等.光谱学与光谱分析),2010,30(3):706.
[11]TAN K,DU P J.Spectroscopy and Spectral Analysis(谭琨,杜培军.光谱学与光谱分析),2008,(9):2009.
[12]HE X L,LIU W H,WANG J F.Journal of Light Scattering(何欣龙,刘文浩,王继芬.光散射学报),2018,30(1):70.