Chemical composition and identification of glass products based on generalized linear model

• Published in: MATEC Web of Conferences Vol. 382; p. 1033
• Main Author: Pan, Zhuo
• Format: Journal Article Conference Proceeding
• Language: English
• Published: Les Ulis EDP Sciences 2023
• Subjects: Barium; Chemical composition; Color; Generalized linear models; glass weathering; high potassium; Iron oxides; lead barium; linear regression model; Perturbation; Potassium; Potassium oxides; Regression models; Standard error; Statistical models; Weathering
• ISSN: 2261-236X; 2274-7214; 2261-236X
• DOI: 10.1051/matecconf/202338201033

Weathering will cause the proportion of chemical composition of ancient glass products to change, thereby affecting the classification judgment of glass type. In order to reasonably analyze the composition content of ancient glass products, this paper provides a reference for studying the content change trend of glass weathering by exploring whether the appearance of glass cultural relics affects the weathering trend and the weathering composition law. According to the relationship between glass surface weathering and ornamentation, color and glass type, a linear regression model is constructed, and in order to solve the problem of heteroscedasticity in the perturbation term, the OLS robust + standard error method is used to improve the model parameters. The results show that the ornaments are B and C and are more likely to weather, high potassium glass is easy to weather, the color is green is mostly unweathered glass, and the color is black There is a tendency to change to weathered glass. The lower the content of potassium oxide, iron oxide and alumina in high potassium, the easier it is to weather, the higher the silica content, the easier it is to weather, and the regularity of weathering and component content of lead barium compared with high potassium is not obvious.