Research and Analysis of Glass Weathering Based on Ergodic K-Prototype Algorithm

Abstract

When making glass products, adding fluxes with different chemical compositions can form different glass types. Affected by the burial environment, the composition proportion of ancient glass products will change over time, thus affecting the correct judgment of their types. Based on decision tree, traversal K-prototype algorithm, grey correlation analysis and other methods, this paper studies and analyzes the classification and identification related to glass weathering and the statistics of chemical composition laws, the relationship between the surface weathering of glass relics and the glass type, decoration and color is analyzed. In this paper, after data pretreatment, the surface weathering and the other three indicators are preliminarily observed through contingency analysis, and then the correlation between the variables is further clarified by using the chi-square test. It is found that the surface weathering is significantly related to the overall glass type, and lead barium glass is more prone to surface weathering than high potassium glass; There is no obvious correlation between surface weathering and decoration and color. Then, according to the glass types, the glass relics are divided into four glass types: lead-barium weathering/non-weathering, high-potassium weathering/non-weathering, and the statistical analysis is established after the pretreatment of the chemical composition of these four types. Then, combined with the independence t-test, we can further obtain the specific element differences, and give a ranking table of the degree of influence. Finally, based on the above statistical exploration, the model of chemical composition law of glass relics before weathering is built according to the mechanism analysis of the four composition laws of high-potassium glass Si and its chemical composition that significantly affects the weathering, the chemical composition that significantly affects the weathering of lead-barium glass, and the chemical composition that has no effect on the weathering of glass. According to the test data of weathering point, the prediction table of chemical composition content before weathering is given. The interval test of the model results shows that the failure rate of high potassium prediction is high. The revised model of chemical composition law of glass relics before weathering is obtained by analyzing and considering the modification of the value rule of high potassium Si.