Composition Analysis and Identification of Ancient Glass Products

Abstract

Based on the research on the rules of the classification of the two kinds of glass as the goal, take glass types as the dependent variable, the chemical composition content is the independent variable, and establish a model of decision tree classification, is based on chemical component content of glass type classification rule, then to analyze the chemical composition of each category, according to the laws of the elbow to calculate the clustering analysis, the optimal class number of k, the K-means clustering algorithm was used to subclassify the glass into K classes and quantify the types. The type was taken as the dependent variable, and the content of each chemical component was taken as the independent variable for decision tree classification. The sub-classification results based on the content of each chemical component and the chemical variables with significant effect on the sub-classification results were obtained. Perturbation was introduced to the chemical variables that had a significant effect on the subclassification results, and the subclassification changes after perturbation were studied to verify the sensitivity of the classification results. The results showed that the accuracy and sensitivity of the model were good.