Abstract: The analysis of the composition of ancient glass on the Silk Road is helpful to understand the context of history. Due to the influence of adverse environmental conditions, the glass will be weathered, which brings difficulties to the composition analysis and classification of glass. In this paper, a comprehensive identification method is proposed, the glass composition before weathering is predicted, and the classification model is established by using voting method and K-Means method. After validation, the category classification model achieves an accuracy of 98.44% with good robustness.