古代玻璃制品成分分析和鉴定的研究

Study on Composition Analysis and Identification of Ancient Glass Products

摘要: 古代玻璃种类繁多且易受环境影响而风化，因此需要对古代玻璃制品的化学成分数据分析，研究有无风化玻璃制品成分的变化规律，以并探索亚分类方法，进而可以根据未知分类的文物化学成分对文物进行准确的分类。本文通过使用K-means算法和BP神经网络结合的方式对玻璃制品进行亚分类划分，之后根据亚分类种类进行风化前后成分的预测；通过RUSBoost机械学习算法，70%的数据作为训练集，15%的数据作为测试集，其余部分作为预测集，来进行玻璃制品的种类鉴定。这些模型相互之间配合紧密，所得结果依次递进，使最终求解真实可靠。模型充分联系实际，具有很好的通用性和推广性。

Abstract: There are many kinds of ancient glass and it is easy to be weathered by the environment. Therefore, it is necessary to analyze the chemical composition data of ancient glass products, study the chang-ing rules of the composition of weathered glass products, and explore subclassification methods, so as to accurately classify cultural relics according to the chemical composition of cultural relics un-known. This paper uses K-means algorithm and BP neural network to subclassify glass products, and then predicts the composition before and after weathering according to the subclassification types. Through the RUSBoost mechanical learning algorithm, 70% of the data is used as the training set, 15% of the data is used as the test set, and the rest of the data is used as the prediction set to identify the types of glass products. The classification rules and identification of ancient glass prod-ucts are divided. These models cooperate closely with each other, and the results are progressive in turn, which makes the final solution true and reliable. The model is fully connected with practice and has good generality and popularization.