Glass Classification Method Based on K-Means++ Algorithm

Abstract

Ancient glass is highly susceptible to weathering by the influence of the buried environment, resulting in changes in the proportion of its components. In order to select the appropriate chemical components for each category and divide them into subclasses, and give specific classification methods and classification results, this paper considers first using the K-means++ algorithm for clustering, and calculating the aggregation coefficient of each k value to determine the number of clusters. The three-level nomenclature is used to name each type according to the amount of chemical component content contained in each category, lists its division basis and the names of each type of high potassium glass and lead barium glass, and then conducts rationality analysis, uses the contour coefficient to evaluate the quality of the cluster, and concludes that the number of clusters of high potassium glass and lead barium glass is 3 and 5 respectively. Finally, sensitivity analysis is performed on the aggregation coefficients.