About this Abstract |
| Meeting |
2027 TMS Annual Meeting & Exhibition
|
| Symposium
|
AI-Enabled Materials Processing: Integrating Accelerated Experimental Workflows and Processing-Aware Machine Learning
|
| Presentation Title |
A Deep Learning-Based Framework for Three-Dimensional Temperature Field Prediction and Intelligent Secondary Cooling Optimization in Continuous Casting |
| Author(s) |
Pan Sun, Yanming Zhang, Wuguo Chen, Xin Xiao, Danbin Jia, Songyuan Ai, Anpu Chen, Mujun Long |
| On-Site Speaker (Planned) |
Pan Sun |
| Abstract Scope |
Efficient secondary cooling optimization is essential for maintaining solidification behavior in continuous casting. Conventional methods rely on repeated calculations of solidification heat-transfer models, resulting in high computational costs. This study developed a deep learning-based framework for real-time 3D temperature field prediction and secondary cooling optimization. A deep neural network (DNN) was trained using temperature field data under various casting conditions, with Bayesian optimization optimizing the network architecture and training parameters. The optimized model achieved a MAE of 0.641 ℃ and an MAPE of 0.062% on the test set with the Swish activation function. The DNN was further employed as a surrogate model for cooling water optimization using five intelligent algorithms. The particle swarm optimization showed the best performance, maintaining the metallurgical length unchanged while achieving a mean absolute shell thickness deviation of 0.12 mm along the casting direction. This framework enables rapid temperature field prediction and efficient secondary cooling optimization. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Machine Learning, Process Technology, Iron and Steel |