摘要
为了建立一种快速无损的手机贴膜检验的分析方法,采用傅立叶变换红外光谱仪对65个手机贴膜样品进行检验。对原始数据进行标准化处理,然后计算二阶导数,最后结合偏最小二乘法降低二阶求导数据的维度。使用聚类分析对样本进行分类,并使用人工神经网络进行分析和预测。结果显示,65个手机贴膜在人工神经网络的训练集和测试集中的正确率均达到97.9%。该方法方便快捷,对样品无损且用量少,为手机贴膜的分类提供有力的支持。
A Fourier transform infrared spectrometer was utilized to analyze 65 mobile phone film samples in order to develop a rapid and non-destructive diagnostic approach for screen protectors inspection.Standardize the original data before computing the secondorder derivative and combining partial least squares to lower the dimensionality of the second-order derivative data.To categorize data,utilize clustering analysis,and for analysis and prediction,use artificial neural networks.The accuracy of 65 screen protectors in both the training and testing sets of the artificial neural network achieved 97.9%,according to the data.This approach is simple and quick,with no sample damage and low consumption,and it provides significant support for the categorization of screen protectors.
作者
姜红
周飞翔
周贯旭
JIANG Hong;ZHOU Fei-xiang;ZHOU Guan-xu(Criminal Investigation Department of Gansu Police Vocational College,Lanzhou 730046,China)
出处
《化学研究与应用》
CAS
北大核心
2024年第4期766-770,共5页
Chemical Research and Application
基金
食品药品安全防范山西省重点实验室开放课题(202204010931006)资助。
关键词
手机贴膜
红外光谱
偏最小二乘法
人工神经网络
screen protectors
infrared spectroscopy
partial least squares method
artificial neural network
