Abstract
Optimization of low-temperature sealing materials is of great significance to improving low-temperature performance and durability. This study leverages DeepSeek-v3 (DS) and GPT-generated data and applies machine learning methods, including XGBoost and neural networks, to perform 3D prediction and analysis of key properties of low temperature sealing materials. Data expansion techniques were employed to enhance data quality and improve model prediction accuracy. Additionally, the study evaluates the applicability of AI-generated data in material performance prediction. The results demonstrate the effectiveness of machine learning in material optimization and provide valuable insights for future optimization strategies.