BridgeNet
收藏资源简介:
BridgeNet是由慕尼黑工业大学团队开发的图结构桥梁数据集,包含20,000个通过组合平衡建模(CEM)方法生成的形态多样的桥梁结构。每个数据点包含三种模态:铰接平衡线框模型、基于内力信息的3D体积网格,以及两个标准视角的渲染图像。数据集通过参数化生成流程构建,并应用几何比例和力学约束进行过滤,确保了数据的工程合理性。该数据集支持图机器学习、多模态学习等任务,旨在解决结构工程领域缺乏公开基准数据的问题,推动概念设计阶段的生成式AI应用和结构分析替代模型开发。
BridgeNet is a graph-structured bridge dataset developed by the team from the Technical University of Munich, containing 20,000 diverse bridge structures generated via the Combined Equilibrium Modeling (CEM) method. Each data point includes three modalities: articulated equilibrium wireframe models, 3D volumetric grids based on internal force information, and rendered images from two standard viewing angles. The dataset is built through a parametric generation pipeline, and filtered using geometric proportional constraints and mechanical constraints to ensure the engineering rationality of all included data. This dataset supports tasks such as graph machine learning and multimodal learning, aiming to address the shortage of publicly available benchmark datasets in the field of structural engineering, and promote the development of generative AI applications and surrogate models for structural analysis during the conceptual design phase.
BridgeNet 数据集概述
数据集基本信息
- 数据集名称:BridgeNet
- 许可证:MIT
- 数据规模:10K < n < 100K(具体为20,000个样本)
- 数据类型:桥梁结构模型数据集
数据集内容
该数据集包含20,000个形态找形桥梁结构,每个桥梁结构包含以下内容:
- 铰接平衡线框模型:使用组合平衡建模(CEM)形态找形方法生成的铰接平衡线框模型。
- 体积化3D网格:通过力信息材料化获得的体积化3D网格。
- 渲染图像:从两个典型摄像机角度渲染的图像。
相关论文
更多关于此数据集的信息可在BridgeNet会议论文中找到。
- 论文标题:BridgeNet: A Dataset of Graph-based Bridge Structural Models for Machine Learning Applications
- 作者:Lazlo Bleker, Mustafa Cem Güneş, Pierluigi DAcunto
- 年份:2025
- arXiv ID:2512.14496
- 论文链接:https://arxiv.org/abs/2512.14496
- 主要分类:cs.CE

- 1BridgeNet: A Dataset of Graph-based Bridge Structural Models for Machine Learning Applications慕尼黑工业大学 · 2025年



