A dataset for machine learning research in the field of stress analyses of mechanical structures

The dataset is prepared and intended as a data source for development of a stress analysis method based on machine learning. It consists of finite element stress analyses of randomly generated mechanical structures. The dataset contains more than 270,794 pairs of stress analyses images (von Mises stress) of randomly generated 2D structures with predefined thickness and material properties. All the structures are fixed at their bottom edges and loaded with gravity force only. See PREVIEW directory with some examples. The zip file contains all the files in the dataset.

本数据集经筹备构建，旨在作为基于机器学习的应力分析方法开发的数据源。其收录了随机生成机械结构的有限元应力分析（finite element stress analysis）结果：数据集包含超过270794组成对的应力分析图像（对应冯·米塞斯（von Mises）应力），这些图像针对预先设定了厚度与材料属性的随机生成二维（2D）结构。所有结构均固定于底部边缘，且仅受重力载荷作用。可查看PREVIEW目录获取部分示例。该ZIP文件包含数据集中的全部文件。