基于随机森林分类模型的古代玻璃制品成分分析

Composition Analysis of Ancient Glass Products Based on Random Forest Classification Model

[目的]中国古代利用外来玻璃制作工艺制作出的玻璃制品,其化学成分与外界传入的并不相同。古代玻璃易受环境影响而被风化,内部元素和外部元素进行大量交换,成分比例会因此发生变化。通过分析玻璃类型等外部因素与表面风化的相关性及玻璃组成成分与表面风化的相关性,找出玻璃表面风化原因。[方法]通过分析玻璃外部因素的相关性和对玻璃进行独立样本Mann-Whitney检验,得到玻璃制品的成分,基于随机森林建立模型,构建不同分类标准。[结果]研究发现,玻璃制品纹饰等与表面风化的相关性较弱,与二氧化硅等玻璃制品的相关性大,且氧化铅成分在玻璃中的重要性最大。[结论]要防止玻璃发生风化,就要减少杂质成分。

[Purposes]The chemical composition of glass products produced by absorbing foreign glass production processes in ancient China is not the same as that imported from the outside world.Ancient glass is susceptible to environmental impact and is prone to being weathered.The internal elements and external elements are exchanged in large quantities,and the composition ratio will change accordingly.By analyzing the correlation between external factors such as glass type and surface weathering and the correlation between glass composition and surface weathering,the reasons for glass surface weathering are found.[Methods]Firstly,the correlation analysis of external factors of glass was carried out,and then the independent sample Mann-Whitney test analysis was carried out to obtain the composition of glass products.Based on the random forest model,different classification criteria were constructed.[Findings]It is found that the correlation between the decoration of glass products and surface weathering is weak,and the correlation with glass products such as silica is strong,and the lead oxide composition is the most important influence factor in glass.[Conclusions]The impurity components should be reduced to prevent the glass from weathering.