Chemical composition and identification of glass products based on generalized linear model

Weathering will cause the proportion of chemical composition ancient glass products to change, thereby affecting classification judgment type. In order reasonably analyze content products, this paper provides a reference for studying change trend weathering by exploring whether appearance cultural relics affects and law. According relationship between surface ornamentation, color type, linear regression model is constructed, in solve problem heteroscedasticity perturbation term, OLS robust + standard error method used improve parameters. The results show that ornaments are B C more likely weather, high potassium easy green mostly unweathered glass, black There tendency weathered glass. lower oxide, iron oxide alumina potassium, easier it higher silica content, regularity component lead barium compared with not obvious.