The weathering of ancient glass changes the amount of its chemical components, influencing the classification. The impact of several elements on weathering is investigated in this research using controlled variables. The composition of numerous glass-like cultural artifacts before and after weathering was examined and forecasted using applicable mathematical statistical information. The Linear SVM method is used to classify and identify glass artifacts. To address the following problems, it is important to analyze and model using the appropriate data provided in the attachment: 1: Examine the link between the glass decoration, kind, and color of these glass objects and their surface deterioration. In addition to the glass type, the statistical law of the content of chemical components that have weathered and not weathered on the surface of samples of cultural artifacts was examined separately, their relationship was discussed, and the chemical composition content prior to weathering was predicted using the data from weathering point detection. 2: The categorization law of lead-barium glass and high-potassium glass is examined in light of the attachment dat. The proper chemical components are chosen to split each category into its subcategories, and particular classification findings and procedures must be provided. The logic and sensitivity of the classification results are also examined.3: In Annex Form 3, analyze the chemical composition of unknown glass artifacts, identify the kind of glass to which they belong, and do sensitivity analysis on the classification results. 4: Analyze the correlation between the chemical compositions of different types of glass cultural relics samples, and examine the differences in the correlation between different categories. Use the strategy of controlling variables to start with for issue 1, merely changing the items in the ornamentation, type, and color each time. Then, consider how these changes affect weathering and determine the connection between the three and the surface of the cultural artifacts. The samples are then separated into four groups according to the sampling site and the category of cultural relics. The chemical components of each type of sample are used as a box-type diagram to eliminate the group value, and then the average value is utilized as the proportional law of this type of sample. Based on the data change before and after we weathering, the effect of the weathering composition of the weathered, it can forecast the content of each component before the sampling point based on this foundation to identify the data based on the realization point. So the solution to issue 1 is as follows. The following factors have an impact on weathering (beginning with the easiest): lines B> C> A; lead barium> high potassium; black = blue green> others in color. For the chemical composition rules of cultural relics and the preliminary forecast of weathering spots, the K2O concentration of high potassium glass is normally approximately 9.3% before weathering, and the weathering has reduced its content to 0.55%. The prediction effect is achieved by 8.8%, and the entire findings are included in the model solution. Based on the successful prediction results of problem 1, the weathered data is replaced with prediction data plus uncontrolled data for problem 2 and the Linear SVM method is used. There are two types of categorization, with 97.2% accuracy in the training set and 100% accuracy in the test set. The same process is used to separate the two types of glass into two subcategories: magnesium and silicon. MGO has a total training set and an accuracy rate of 96.6%, a total test set and an accuracy rate of 100%, and a decent classification. The data is then disrupted by the sensitivity test. When the disturbance range approaches 20%, the overall accuracy rate is 88%, and the sensitivity is low, indicating that the stability is improved.In response to problem 3, the model of problem 2 is categorized, and the data is disturbed based on the classification findings, progressively increasing the range of disturbances to 20%. There is no change, suggesting that the sensitivity is low and the stability is high. The categorization findings are shown, such as polarized high potassium glass in the A1 cultural relics category and silicon lead glass in the A2 cultural relics category. The model solution contains the remaining outcomes. In response to problem 4, we use the Spearman correlation coefficient to examine the relationship between the chemical components of high potassium and lead-barium, and we calculate the P value using the p value. Different studies of the two varieties of glass of the same component P value differences between the two. The particular correlation coefficient findings are presented in the model solution, and the difference between the P values is that the results of the K2O and CAO in the high potassium glass are the strongest, with a P value of 0.000496, and the MGO and AL2O3 are significant in the lead glass glass. The highest P value is 0.00000189. Finally, we assessed the model’s merits and weaknesses, and the model was promoted to other fields of archeology and history on the basis of suggesting comparable improvement solutions for the flaws.