摘要详情
12 / 2025-02-28 13:55:52
A Data-Model Synergetic Framework Integrating Machine Learning with S-MDMS for Digital Process Optimization
thermo physical parameter, machine learning, forecasting model, database construction
摘要待审
Under the accelerating advancement of intelligent manufacturing, precise control of steel material heat treatment processes faces dual challenges: insufficient accuracy in traditional empirical models and low computational efficiency of finite element methods. This study develops a deep-integration architecture combining machine learning prediction modules with a Steel Material Data Management System (S-MDMS), achieving synergistic optimization between intelligent thermal property prediction and full lifecycle data management. Within the S-MDMS framework, we constructed a multidimensional database encompassing material composition, phase transformation characteristics, process history, and service performance. We implemented ontology-based data modeling for the semantic fusion of multi-source heterogeneous data. We innovatively established a machine learning-based multiscale prediction system to overcome the limitations of conventional finite element methods that rely on empirical constitutive equations. By systematically integrating random forests, gradient-boosting decision trees, and XGBoost algorithms, we developed quantitative mappings between material composition, microstructure, process parameters, and thermal properties. Experimental results demonstrate the superior performance of XGBoost, with Bayesian-optimized hyperparameters yielding an ensemble model achieving R² > 0.99 prediction accuracy on test datasets. The proposed three-dimensional "data governance-feature extraction-predictive feedback" collaborative framework aligns with computational materials engineering principles in metallurgy, providing an innovative paradigm for the digital transformation of heat treatment processes in intelligent manufacturing ecosystems.
重要日期
-
会议日期
08月19日
2025
至08月22日
2025
-
06月30日 2025
初稿截稿日期
-
08月22日 2025
注册截止日期
主办单位
中国机械工程学会热处理分会
协办单位
《材料热处理学报》
《金属热处理》
《金属热处理》
联系方式
- IFHTSE 2025 OFFICE
- ch******@chts.org.cn
- 010*********
