A fully GPU-based workflow for building physics emulators of hypersonic flows
作者: Fabian Paischer, Dylan Rubini, Deniz A. Bezgin, Aaron B. Buhendwa, David Hauser, Florian Sestak, Johannes Brandstetter, Sebastian Kaltenbach, Nikolaus A. Adams
分类: cs.LG, cs.AI, physics.comp-ph, physics.flu-dyn, stat.ML
发布日期: 2026-06-11
备注: First authors contributed equally
💡 一句话要点
提出全GPU工作流程以构建高超音速流的物理仿真器
🎯 匹配领域: 支柱八:物理动画 (Physics-based Animation)
关键词: 高超音速流动 物理仿真 神经网络 GPU加速 数据生成 残差改进 工程设计
📋 核心要点
- 高超音速流动的复杂物理现象难以通过传统模型准确预测,尤其是冲击波位置和强度的捕捉存在挑战。
- 本文提出了一种全GPU工作流程,结合加速数据生成与神经网络训练,增强物理一致性和不确定性量化。
- 通过残差改进,仿真器在仅有网格和输入参数的情况下进行训练,显著降低残差并提高物理一致性。
📝 摘要(中文)
解决复杂物理现象的高保真度与低计算成本是现代工程中的关键挑战,尤其是在高超音速流动中,精确预测流场拓扑至关重要。传统的降阶模型和神经仿真器在捕捉流态的陡峭梯度时存在困难。为此,本文提出了一种全GPU工作流程,结合加速数据生成与神经仿真器的训练,增强物理一致性。该工作流程基于可微分的高保真求解器(JAX-Fluids),用于快速数据集创建和基于残差的改进,确保仿真器在训练分布之外的可靠性,满足实际工程设计的需求。
🔬 方法详解
问题定义:本文旨在解决高超音速流动中传统模型无法准确捕捉流态陡峭梯度的问题,尤其是在工业应用中的物理一致性不足。
核心思路:提出一种全GPU的工作流程,利用可微分的高保真求解器生成数据,并通过残差改进增强神经仿真器的物理一致性。
技术框架:整体架构包括数据生成模块、神经仿真器训练模块和残差改进模块,形成一个闭环的训练与优化流程。
关键创新:最重要的创新在于结合了可微分求解器与残差改进技术,使得仿真器能够在训练分布之外保持可靠性,这是传统方法所无法实现的。
关键设计:在网络结构上,采用多种模型架构进行比较,设置了特定的损失函数以优化物理一致性,并通过残差反馈机制进行动态调整。
🖼️ 关键图片
📊 实验亮点
实验结果表明,采用残差改进后,仿真器在仅有网格和输入参数的情况下,残差显著降低,物理一致性提升,验证了该方法在高超音速流动预测中的有效性,超越了传统模型的性能。
🎯 应用场景
该研究的潜在应用领域包括航空航天、汽车工程和气动设计等高超音速流动相关的工程问题。通过提供高效的物理仿真工具,能够加速设计过程,降低开发成本,并提高产品性能,具有重要的实际价值和广泛的未来影响。
📄 摘要(原文)
The ability to resolve complex physical phenomena with high fidelity and at low computational cost is central to addressing key challenges in modern engineering. A prime example lies in hypersonic flows, where the precise prediction of the full flowfield topology, in particular with respect to shock wave location and intensity, is critical. Yet supersonic and hypersonic flows continue to be a stumbling block for traditional reduced-order models and neural emulators that struggle to capture steep gradients in flow states with physical consistency in applications of industrial relevance. To that end, we introduce a fully GPU based workflow that integrates accelerated data generation with the training of neural emulators augmented by uncertainty quantification and physics-aware refinement. Our workflow is enabled by a differentiable high-fidelity solver (JAX-Fluids) which we employ for rapid dataset creation and residual-based improvement of the neural emulator to enhance physical consistency. Building on this framework, we first present a suite of model architectures and analyze their scaling behavior to expose their strengths and shortcomings. We then show that residual-based refinement enables training on cases where only mesh and input parameters are available, substantially reducing residuals and improving physical consistency. Together, differentiable simulation and residual-based refinement yield physics emulators that remain reliable beyond their training distribution, a key requirement for deploying surrogates in real-world engineering design loops.