Classification and identification of glass based on principal component and systematic cluster analysis

Abstract

The composition of glass relics in different regions is not completely the same. Meanwhile, due to the large amount of internal elements exchanged in the process of glass burial and weathering, in order to determine and find a method to classify and identify glass relics, principal component analysis and systematic clustering are used to select the appropriate indexes for clustering by selecting the composition indexes of existing glass samples. The types of unknown glass artifacts are determined by classification from the types of known glass artifacts. Finally, after multiple cluster analysis, two unknown types of glass relics were identified as high potassium and lead barium. After the identification, the glass chemical composition data were fine-tuned up and down, and it was found that the identification type remained unchanged and the identification effect was stable and accurate.