距离感知注意力与壁面距离专家路由用于基于Transformer的三维流场预测
Distance-Aware Attention and Wall-Distance Expert Routing for Transformer-Based 3D Flow Prediction
浏览论文内容
中文总结 AI 辅助
提出距离感知交叉注意力和表面-体积混合专家路由，基于壁面距离条件化Transformer代理模型，显著提升三维流场预测精度。
中文摘要 AI 辅助
用于三维流场预测的Transformer代理模型将工业网格压缩为一小组令牌，每个预测点从这些令牌中读取信息。随后执行两个操作：检索步骤（点从压缩表示中收集信息）和前馈层（转换所检索到的信息）。在当前的主干网络中，这两个操作都对点在流场中的位置不敏感。我们将两者都基于与壁面相关的物理信号进行条件化。距离感知交叉注意力（DA-CA）在检索前根据壁面距离重塑每个体积查询，使得边界层深处的点与外部流中的点提取不同的几何信息。表面-体积混合专家（SVMoE）用一小组专家替换共享的前馈层，对于体积点根据壁面距离路由，对于表面点根据局部几何路由。这两种机制都不依赖于特定架构，因此我们将它们原封不动地应用于AB-UPT和Transolver-3。在包含50个训练案例的DrivAerML上，DA-CA将体积压力误差降低了10.1%，DA-CA和SVMoE共同将其降低了12.5%；DA-CA改善了近壁区域，但以远场区域为代价，SVMoE恢复了这一损失，并且体积专家在没有路由监督的情况下自动划分为近壁、过渡和自由流区域。在300个案例上重新训练后，条件化改进了所有场量，在AB-UPT上将体积压力和速度误差分别降低了33.1%和18.6%，在Transolver-3上分别降低了21.4%和21.3%。在DrivAerNet++上的留一车身评估中，它将未见车身类型的体积压力误差最多降低了14.2%。
英文摘要
Transformer surrogates for 3D flow prediction compress an industrial mesh into a small set of tokens from which every prediction point reads. Two operations follow: the retrieval step in which a point gathers information from the compressed representation, and the feed-forward layer that transforms what it retrieved. In current backbones both are blind to where the point sits in the flow. We condition both on wall-related physical signals. Distance-aware cross-attention (DA-CA) reshapes each volume query by its wall distance before retrieval, so that a point deep in the boundary layer draws different geometric information than one in the outer flow. Surface-volume mixture-of-experts (SVMoE) replaces the shared feed-forward layer with a small set of experts, routed by wall distance for volume points and by local geometry for surface points. Neither mechanism is tied to one architecture, so we apply both unchanged to AB-UPT and Transolver-3. On DrivAerML with 50 training cases, DA-CA reduces the volume pressure error by 10.1%, and DA-CA and SVMoE together reduce it by 12.5%; DA-CA improves the near-wall region at some cost in the far region, which SVMoE recovers, and the volume experts settle into near-wall, transition, and free-stream bands without routing supervision. Retrained on 300 cases, the conditioning improves every field quantity, reducing volume pressure and velocity errors by 33.1% and 18.6% on AB-UPT and by 21.4% and 21.3% on Transolver-3. Under Leave-One-Body-Out evaluation on DrivAerNet++, it reduces the volume pressure error on unseen body types by up to 14.2%.
发表机构
- Korea Advanced Institute of Science and Technology（韩国科学技术院）
- Hanyang University（汉阳大学）
- Narnia Labs（纳尼亚实验室）
机构由 AI 辅助整理，请以论文原文为准。
