Ancient glass products are very vulnerable to weathering caused by burial environment. In the process of weathering, a series of changes will often occur in the proportion of chemical components inside. The chemical composition of these glass relics was analyzed and identified.This paper makes chi square test and Spearman correlation analysis on the surface weathering of glass cultural relics and other attribute values respectively. It is found that the surface weathering of cultural relics is closely related to the type of cultural relics and has significant differences. The quantitative relationship between the weathering of cultural relics and other attribute values is fitted through multiple regression in three ways. Finally, pure quadratic multiple regression is selected, and the fitting accuracy reaches 75.9259%. Then, the data are normalized, and the rule of whether there is weathering chemical composition content in cultural relics samples is summarized through the relationship diagram between the content of each chemical component and the type of chemical component in each sample. Then, the influence of each component on the weathering results of the samples is obtained by using the principal component analysis method to verify the rationality of the conclusion. It is found that the content of phosphorus pentoxide and copper oxide has a significant impact on the weathering results of cultural relics. Finally, given the initial values of the variables in the weathering samples that have a low impact on the weathering degree, the chemical composition content of the samples before weathering can be predicted by multiple linear regression.In this paper, firstly, based on the two types of high potassium glass and lead barium glass, the average and variance of the chemical composition content in each sample were counted, and the classification rules of the two types of glass were obtained by analyzing the difference of the chemical composition of the two types of glass. Then, using the Kmeans clustering method, the high potassium glass and lead barium glass were clustered, and the principal component analysis method was used to reduce the dimensions of the data, Combined with each principal component and the clustering process, the sub classification standard was determined. Finally, the sensitivity of the classification results was analyzed by PSI method, and the rationality of the classification results was analyzed by sampling inspection method. The accuracy of the classification results was 89.036%.

Qing Li, Chuang Wang, Jianwei Zhai. Research on Composition Analysis and Identification Based on Ancient Glass Products. Academic Journal of Materials & Chemistry (2022) Vol. 3, Issue 2: 63-72. https://doi.org/10.25236/AJMC.2022.030210.

[1] Gan Fuxi, Zhao Hongxia, Li Qinghui, Li Ling, Cheng Huansheng. Scientific and technological analysis and research on glass products unearthed in the Warring States Period in Hubei Province [J]. Jianghan Archaeology, 2010 (02): 108-116+151+0

[2] Wang Chengyu, Jiang Yanyan, Pang Shihong, Tang Huajuan, Tao Ying. Development of glass composition design[J]. Journal of the Chinese Ceramic Society, 2007, 35:143-148

[3] Fang Xiangzhong. Chi Square Distribution and Chi Square Test [J]. China Statistics, 2022 (05): 29-31

[4] Wang Xiaoyan, Li Meimei. A Brief Talk on Grade Correlation Coefficient and Spearman Grade Correlation Coefficient [J]. Journal of Guangdong Light Industry Vocational and Technical College, 2006, 5 (4): 2

[5] Qu Junhua, Ni Jiaming. Analysis and design of multiple regression model [J]. China Electric Power Education, 2007

[6] Jolliffe I T. Principal Component Analysis Springer Verlag [J]. Proceedings of wseas int.conf.on signal speech & image processing corfu, 2002.

[7] Liu Hailin, Xiao Junrong. An evolutionary super multi-objective optimization dimension reduction algorithm based on decomposition and hyperplane fitting [J]. Journal of Electronics and Information, 2022,44 (09): 3289-3298

[8] Wang Songgui. Linear statistical model: linear regression and analysis of variance [M]. Higher Education Press, 1999