A study of glass classification and identification based on K-means clustering and Euclidean distance

Abstract

Glass is buried in the process of weathering, resulting in a large number of exchanges of its chemical composition. To classify and identify the glass of unknown composition, firstly, the existing classification law of glass was analyzed, and the subclasses of glass types were divided based on principal component analysis and K-means cluster analysis, and the number of subclasses was determined according to the P value of cluster significance, and the high potassium glass was divided into 3 classes and the lead-barium glass was divided into 4 classes. Subsequently, a glass classification model was constructed for the unknown component glasses, and after obtaining the glass classification, the classification model of glass mean was constructed by Euclidean distance based on K-means clustering results, and the subclass classification results of all unknown samples were obtained.