Two-dimensional temperature field prediction of rotary kiln based on graph neural networks

Physics of Fluids, Год журнала: 2025, Номер 37(2)

Опубликована: Фев. 1, 2025

Sensing and optimizing the temperature distribution in rotary kilns is key to improve energy efficiency reduce production costs. Traditional computational fluid dynamics (CFD) solvers are computationally expensive cannot meet demand for real-time performance industrial sites. With continuous development of deep learning, graph neural networks (GNNs) have emerged as a potentially effective method accelerating CFD unstructured grid simulations. In order accurately predict whole field kiln, novel GNN model designed this study, CLJPNet proposed fast prediction kiln. Compared with traditional GNN, study able kiln by using Cleary-Luby-Jones-Plassmann Coarsening coarsening algorithm multi-algebraic lattice sparsify topology accelerate inference speed while maintaining high accuracy. Finally, paper compared other three models verify effectiveness model. The experimental results indicate that achieves coefficient determination (R2) 0.99, mean squared error 710.63, absolute percentage 1.64, relative region interest 0.02 on test set, all evaluation metrics superior models, demonstrating better performance. addition, runs 3 orders magnitude faster than rapid fields provides approach intelligent advancement production.

Язык: Английский

---

Advancing understanding of indoor conditions using artificial intelligence methods

Physics of Fluids, Год журнала: 2025, Номер 37(1)

Опубликована: Янв. 1, 2025

This study presents a novel methodology for optimizing probe placement in indoor air-conditioned environments by integrating computational fluid dynamics simulations with artificial intelligence techniques an unsupervised learning framework. The “Reduce Uncertainty and Increase Confidence” algorithm identified spatially distinct thermal velocity clusters based on temperature magnitude distributions. Optimization of positions within these clusters, guided sequential least squares programing, resulted effective strategy to minimize redundancy while maximizing spatial coverage. highlights the interplay between temperature, relative humidity, velocity, turbulence intensity, revealing critical insights into airflow behavior its implications occupant comfort. findings presented underscore potential targeted provide robust framework advanced climate control.

Язык: Английский

---

Transforming Hospital HVAC Design with BIM and Digital Twins: Addressing Real-Time Use Changes

Sustainability, Год журнала: 2025, Номер 17(8), С. 3312 - 3312

Опубликована: Апрель 8, 2025

Traditional HVAC designs often struggle to respond promptly and accurately dynamic changes in complex environments like hospital usage. This paper introduces a novel framework that integrates Building Information Modeling (BIM), digital twin technology, practical medical processes transform design for construction. The ensured smarter (with reduction of 90% calculation time an improvement 38.20–53.24% respondence speed) cleaner environment after identifying calculating the rational layout functional areas optimizing intersecting flow lines. A key innovation this research was application Support Vector Machine (SVM) deep learning algorithm (Long Short-Term Memory) networks real-time pedestrian traffic prediction. implementation validated through multiple simulations applications including horizontal vertical negative pressure analyses three distinct departments. findings underline potential BIM twins optimize systems design, providing adaptive, data-driven solutions both routine operations emergency scenarios. offers scalable approach modernizing healthcare infrastructure, ensuring resilience efficiency diverse operational contexts.

Язык: Английский

---

Inverse design of air supply distribution using coupled adjoint and parametric level set method

Building and Environment, Год журнала: 2025, Номер unknown, С. 113106 - 113106

Опубликована: Апрель 1, 2025

Язык: Английский

---

Applying a physics-informed neural network to an indoor airflow time-extrapolation prediction

Building and Environment, Год журнала: 2025, Номер unknown, С. 113246 - 113246

Опубликована: Май 1, 2025

Язык: Английский

---