基于Fisher判别分析的一次性塑料手套光谱鉴别

Spectrum Identification of Disposable Plastic Gloves Based on Fisher Discriminant Analysis

为了建立一种快速准确且无损检材的检验一次性塑料手套的光谱分析方法,利用X-MET7000手持式X荧光光谱能量色散型分析仪,对39个不同品牌不同用途的一次性塑料手套进行检验分析,通过特征元素Cl与Cd将样品分为4组,对于同组样本还可通过Ca/Zn含量比进行进一步认定,效果较好。使用Fisher判别分析法,建立判别函数Y1、Y2与Y3,实现对新数据类别变量取值的预测,同时检验三个判别函数的判别能力。根据判别能力较强的第一和第二判别函数建立判别分布图,图中4组样本质心明显区分。为验证分类结果的准确性,对所得数据进行系统聚类和K-Means快速聚类,根据数据间的亲疏程度,进行分类,此方法为犯罪现场检验此类物证提供依据。

In order to establish a fast,accurate and nondestructive spectrum analysis method for testing disposable plastic gloves,39 disposable plastic gloves of different brands and applications were tested and analyzed by X-MET7000 handheld X-ray fluorescence spectrum energy dispersive analyzer.The samples were divided into four groups by the characteristic elements chlorine(Cl) and cadmium(Cd).For the same group of samples,the calcium(Ca)/zinc(Zn) content ratios could be used for further analysis and the result was good.Fisher discriminant analysis method was used to establish discriminant functions Y1,Y2 and Y3 to predict the value of new data category variables and test the discriminant ability of the three discriminant functions.According to the first and second discriminant functions with strong discriminant ability,a discriminant distribution map was established and the four groups of sample centroids were clearly distinguished.In order to verify the accuracy of the classification results,the obtained data were clustered systematically and K-means fast,and classified according to the degree of intimacy between the data.This method provided the basis for the crime scene inspection of such physical evidence.