基于成分数据分析与模糊模式识别的古代玻璃种类鉴别  

Identification of Ancient Glass Types Based on Composition Data Analysis and Fuzzy Pattern Recognition

作　　者：王保乾 蒋剑军 WANG Baoqian;JIANG Jianjun

机构地区：[1]铜陵学院数学与计算机学院,安徽铜陵244061

出　　处：《科技创新与应用》2024年第7期41-46,共6页Technology Innovation and Application

基　　金：安徽省重点研究与开发计划项目(202004a05020010);安徽省2020年高校自然科学重点研究项目(KJ2020A0973);安徽省级教学团队(2019jxtd105);安徽省精品线下开放课程(2020kfkc553)。

摘　　要：以一批古代玻璃制品成分数据为例,研究成分数据处理方法与模糊模式识别在古代玻璃分类鉴别中的应用。采用非对称、对称与等距3种对数比变换方法,对古代玻璃成分数据进行转换,针对数据中零值不能进行对数比转换的情况,引入可以保持单形空间运算的乘法替换法。分别对比3种对数比变换方法对模糊模式识别结果的影响,同时引入随机森林与支持向量机算法,对模糊模式识别方法做进一步的分析。近似零值替换后古代玻璃成分数据的统计特征基本不变,乘法替换法具有良好的替换效果。模糊模式识别结果的准确率与变换方法特性表明,基于对称对数比变换的模糊模式识别方法最优。在古代玻璃的分类鉴别研究中引入成分数据分析理论,为古代玻璃成分数据的处理提供一个科学的方法。模糊模式识别方法的引入,为考古工作者、地质研究人员等,提供一种新的简单快速且准确的分类方法。Taking the composition data of a batch of ancient glass products as an example,this paper studies the application of composition data processing method and fuzzy pattern recognition in the classification and identification of ancient glass.Three logarithmic ratio transformation methods of asymmetry,symmetry and equidistance are used to transform the ancient glass composition data.In view of the fact that the zero value in the data can not be transformed by logarithmic ratio,a multiplicative substitution method which can keep the space operation of simplex is introduced.The effects of three logarithmic ratio transformation methods on fuzzy pattern recognition results are compared respectively,and random forest and support vector machine algorithms are introduced to further analyze the fuzzy pattern recognition method.Through the approximate zero value replacement,the statistical characteristics of the ancient glass composition data are basically unchanged,and the multiplication substitution method has a good replacement effect.The accuracy of fuzzy pattern recognition results and the characteristics of transformation methods show that the fuzzy pattern recognition method based on symmetric logarithmic ratio transformation is the best.The theory of composition data analysis is introduced into the classification and identification of ancient glass,which provides a scientific method for the processing of ancient glass composition data.The introduction of fuzzy pattern recognition method provides a new simple,fast and accurate classification method for archaeologists and geological researchers.

关 键 词：古代玻璃 成分数据 零值替换 对数比 模糊模式识别 

分 类 号：K854.2[历史地理—考古学及博物馆学]