Glass composition analysis and identification model based on variance test

Abstract

In this study, a qualitative and quantitative study was conducted on three factors: weathering condition and glass type, decoration, and color; bar graphs and line graphs were drawn and described for each factor at different levels of weathering condition; the relationship between the three factors and weathering condition was analyzed using chi-square test, and the correlation between glass type and weathering condition was obtained to be high; the correlation analysis (2) based on APRIORI algorithm (1) was used to The correlation analysis (2) based on APRIORI algorithm (1) was used to quantitatively calculate the relationship between different levels under each factor and the weathering condition, and it was obtained that the levels of ornamentation B, light blue and blue-green, high potassium and lead-barium had more significant effects on the weathering condition, and the results were consistent with the qualitative analysis. For the second sub-question of question 1, the chemical components with high variability between weathered and unweathered samples were selected using the MannWhitney test, and ridge regression was used to establish a model of weathering condition and these chemical components.For question 2, this paper conducted independent sample MannWhitney tests on glass type and each chemical composition to obtain twelve chemical composition indicators with significant differences between high potassium and lead-barium glasses. Specific functions between glass type and each component were fitted by ridge regression on these chemical components for high potassium and lead-barium glasses.