Reproduce “Application of Zone Method based Physics-Informed Neural Networks in Reheating Furnaces” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Ujjal Kr Dutta, Aldo Lipani, Chuan Wang, Yukun Hu
Researchers from University College London and Swerim AB developed a Physics-Informed Neural Network (PINN) that integrates the zone method's energy balance equations to predict temperatures in reheating furnaces. This approach enables real-time temperature predictions with enhanced accuracy and robustness for industrial process optimization.
Key insights
- Integrating physics-based regularizers significantly enhances a neural network's generalization capability and robustness, particularly in auto-regressive and out-of-distribution prediction scenarios for complex industrial systems.
- Providing previous time-step temperatures as inputs substantially improves the prediction accuracy for all models by offering crucial temporal context.
- The combined use of energy balance regularizers for both gas (volume) and surface zones proved more effective than using either in isolation, confirming the complementary nature of these physical constraints.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.