arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.06275physics.flu-dyncs.LG
收藏

神经SPH:改进的拉格朗日流体动力学神经建模

Neural SPH: Improved Neural Modeling of Lagrangian Fluid Dynamics

Technical University of Munich(慕尼黑工业大学);Munich Institute of Integrated Materials, Energy and Process Engineering, Technical University of Munich(慕尼黑工业大学慕尼黑集成材料、能源与过程工程研究所);Johannes Kepler University(约翰内斯开普勒大学);NXAI GmbH(NXAI有限公司)

Artur P. Toshev, Jonas A. Erbesdobler, Nikolaus A. Adams, Johannes Brandstetter

AI总结:

本文针对基于图神经网络的SPH流体模拟器因拉伸不稳定性导致粒子聚集、难以长期稳定预测的问题,提出将标准SPH求解器中的压力、粘性和外力组件融入GNN训练与推理,显著降低 rollout 误差并提升物理建模能力。

摘要:

光滑粒子流体动力学(SPH)在现代工程和科学学科中无处不在。SPH是一类拉格朗日格式,通过有限的材料点来离散流体动力学,这些材料点通过演化的速度场进行跟踪。由于模拟具有类粒子特性,图神经网络(GNN)已成为颇具吸引力且成功的替代模型。然而,此类基于GNN的模拟器的实际效用依赖于其忠实建模物理的能力,能够在长时间范围内提供准确且稳定的预测——这是一个众所周知的难题。在这项工作中,我们将源自拉伸不稳定性的粒子聚集识别为主要缺陷之一。基于这些见解,我们使用标准SPH求解器中的不同组件(包括压力、粘性和外力组件)来增强最先进的基于GNN的模拟器的训练和 rollout 推理。所有经过神经SPH增强的模拟器均实现了比基线GNN更好的性能,在 rollout 误差方面通常提升数个数量级,从而允许显著更长的 rollout 和显著更好的物理建模。代码可在 https://github.com/tumaer/neuralsph 获取。

英文摘要:

Smoothed particle hydrodynamics (SPH) is omnipresent in modern engineering and scientific disciplines. SPH is a class of Lagrangian schemes that discretize fluid dynamics via finite material points that are tracked through the evolving velocity field. Due to the particle-like nature of the simulation, graph neural networks (GNNs) have emerged as appealing and successful surrogates. However, the practical utility of such GNN-based simulators relies on their ability to faithfully model physics, providing accurate and stable predictions over long time horizons - which is a notoriously hard problem. In this work, we identify particle clustering originating from tensile instabilities as one of the primary pitfalls. Based on these insights, we enhance both training and rollout inference of state-of-the-art GNN-based simulators with varying components from standard SPH solvers, including pressure, viscous, and external force components. All Neural SPH-enhanced simulators achieve better performance than the baseline GNNs, often by orders of magnitude in terms of rollout error, allowing for significantly longer rollouts and significantly better physics modeling. Code available at https://github.com/tumaer/neuralsph.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑