Composition analysis and classification prediction of ancient glass

Abstract

Ancient glass is easily affected by environmental factors, which cause chemical changes between glass composition and environmental substances and leads to weathering. Analysis of the chemical composition of the ancient glass plays an important role in the study of the Silk Road. To study the composition of glass relics, the Kruskal-Wallis test was used to determine the chemical composition of different glass types; A decision tree was established to explore and visualize the classification rules of glass types, and the sensitivity analysis of subclass classification was carried out by setting the value of chemical composition content of cultural relics sampling points from -10% to 10%. It was found that high potash glass was most sensitive to SiO₂, and lead-barium glass was most sensitive to PbO. Finally, based on the random forest algorithm, the sensitivity of the model is analyzed. According to the experimental hypothesis and data validation, it is proved that the comprehensive model has good robustness, and can accurately and quickly analyze and identify the chemical composition of ancient glass.