Composition Analysis and Identification of Ancient Glass Products Based on Optimized BP_Adaboost Strong Classifier

Abstract

Due to the different combustion agents used in the manufacture of glass in ancient China and abroad, there are apparent differences in the composition of ancient glass products in China and abroad. This paper presents a reduction analysis and identification of the composition of ancient glass products, which is of great significance to the study of the Silk Road and the history of the development of glass products. This paper analyzes and identifies the composition of ancient glass products based on the data of C questions of the CUMCM competition. The study first used a neural network optimized by a genetic algorithm to screen out the elements closely related to the weathering condition and used regression to reduce the composition of the weathered samples. The reduced samples were subclassified by systematic clustering according to their respective compositions, and high-potassium glass and lead-barium glass were classified into two subclasses each. The classification basis was reasonably clarified by factor analysis. Finally, the strong classifier was optimized to identify and classify the samples based on the unknown sample components. The correct identification rate reached 100%, which played a good role in distinguishing the weathered Chinese and foreign glass products.