%0 Journal Article
%T 古代玻璃制品的成分分析与鉴别的模型研究<br>Research on the Model of Ancient Glass Products Classification and Identification
%A 刘泊志
%A 李田丰
%A 吕珊珊
%A 刘姝邑
%J Modeling and Simulation
%P 1185-1198
%@ 2324-870X
%D 2023
%I Hans Publishing
%R 10.12677/MOS.2023.122112
%X 为帮助古代玻璃制品鉴别者通过化学成分的定量分析判断玻璃文物类型，解决如何对玻璃制品准确分类的问题，对五十余件玻璃样品多处采样，以采样得到的六十余份数据进行量化分析，使用K均值聚类模型、随机森林模型对其种类进行分析与验证，筛选出对玻璃样品是否风化以及玻璃样品的种类划分有显著性区别的化学成分，建立数学模型并通过将已知数据分为训练集和验证集交叉验证去评价模型准确度，并对灵敏度进行分析，得到判断玻璃制品是否风化以及分类玻璃制品的最优模型。<br />
In order to help archeologist to quantitatively classify the type of relics of glass by the difference of chemical composition and solve the problem of Ancient glass products classification accurately, this paper proposes to make a quantitative analysis of the more than sixty pieces of data obtained from sampling more than 50 ancient glass samples, analyzing and verifying the species by using K-means Clustering model and Random forest model, screening the chemical composition which make a sig-nificant role in classification of whether weathering or type of glass, building a mathematical model, valuing the accuracy of the model by dividing data into two categories of test and validation and an-alyzing its sensitivity, then acquiring the optimal model. The purpose of this study is to help arche-ologist to quantitatively classify the type of relics of glass by the difference of chemical composition.
%K 古代玻璃制品分类，配对t检验，K均值聚类，随机森林，交叉验证；Ancient Glass Products Classification
%K Matched Samples T-Test
%K K-Means Clustering
%K Random Forest
%K Cross-Validation
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=62676