Physics-informed Neural Networks (PINN) for computational solid mechanics: Numerical frameworks and applications

Thin-Walled Structures, Год журнала: 2024, Номер 205, С. 112495 - 112495

Опубликована: Сен. 24, 2024

Язык: Английский

---

Deep-learning-based aesthetic evaluation network for bridge pylon and aesthetics-oriented bridge design

Structures, Год журнала: 2025, Номер 71, С. 108167 - 108167

Опубликована: Янв. 1, 2025

Язык: Английский

---

Uncertainty-oriented dynamic topology optimization for cross-scale concurrent design considering improved size-controlling strategy

Reliability Engineering & System Safety, Год журнала: 2025, Номер unknown, С. 110819 - 110819

Опубликована: Янв. 1, 2025

Язык: Английский

---

Developing optimum topology of mid-plate in cable bracing systems through deep learning

Structures, Год журнала: 2025, Номер 75, С. 108797 - 108797

Опубликована: Апрель 8, 2025

Язык: Английский

---

Two stage multiobjective topology optimization method via SwinUnet with enhanced generalization

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Март 18, 2025

Topology optimization is a critical tool for modern structural design, yet existing methods often prioritize single objectives (e.g., compliance minimization) and suffer from prohibitive computational costs, especially in multi-objective scenarios. To address these limitations, this paper introduces novel two-stage topology (MOTO) method that uniquely integrates data-driven learning with physics-informed refinement, both stages are implemented within nearly identical network frameworks, ensuring simplicity consistency execution. Firstly, MOTO mathematical model based on the constraint programming considers competing of compliance, stress distribution, material usage was constructed. Secondly, neural incorporates shifted windows attention mechanism lightweight modules developed to enhance feature extraction while maintaining efficiency. Finally, proposed trained two stages: In Stage-1, utilizing adaptive input tensors, predicts near-optimal geometries across variable design domains (including rectangular L-shaped configurations) diverse boundary conditions real time, requiring only 1,650 samples per condition. Stage-2, structures Stage-1 were physically optimized achieve optimal performance. Experimental results demonstrate method's capability generate high-accuracy, computationally efficient solutions robust generalization capabilities. It effectively tackles challenges associated multi-scale non-convex geometries, various even untrained significantly reducing data dependency, advancement optimization. The approach offers new insights promotes advancements practices.

Язык: Английский

---

Deep Learning-based Topology Optimization for Multi-Axis Machining

Applied Mathematical Modelling, Год журнала: 2024, Номер 138, С. 115738 - 115738

Опубликована: Окт. 11, 2024

Язык: Английский

---