Automatic defeaturing of CAD geometries for simulation of complex systems

工程师使用的计算机模型的几何形状可以非常详细，特别是在准备制造的情况下。生产级几何可以包含间隙、干涉、紧固件和非常小的特征。这些特性通常是制造所必需的，但会给仿真增加不必要的复杂性。这些细节的存在会增加模拟的计算时间，而在预测的准确性方面不会给工程师带来任何好处。因此，仿真工程师从生产级几何结构中去除特征，这一过程被称为特征化。特征处理通常是手动进行的，因为它需要专业知识。这个项目的目的是研究是否可以使用计算机算法自动执行去特征。目标是：(i)调查用于表示和存储计算机模型、其模拟参数及其特征的方法；（ii）开发用于自动决策的去特征算法，并将其作为新的工程软件实施；（iii）使用选定的案例研究评估开发的工具。学生将开发一个计算环境来实验适当的计算机算法。这将涉及实现模拟工作流以支持选定的案例研究。然后，学生将研究特征和仿真参数表示，定义一个特征分类法，可以为去特征算法的开发提供信息。这项工作将涉及将几何特征与模拟环境联系起来。将开发一种决策工具，根据模拟上下文将特征化算法应用于几何形状。一系列的计算机算法将被测试和比较。通过与工业界专家接触，将获得对其感知有效性的评估。总之，该项目旨在提供一个环境感知特征系统，以提高工程设计的生产力。研究领域：工程设计、几何与拓扑学、图形与可视化

The geometry of computer models used by engineers can be very detailed, especially if they are ready for manufacturing. Production-level geometry can contain gaps, interferences, fasteners, and very small features. These features are often necessary for manufacturing, but can add unnecessary complexity for simulation. The presence of such details can drive up the computational time of simulation, without providing any benefit to the engineer in terms of accuracy of prediction. Therefore, simulation engineers remove features from production-level geometry, a process referred to as defeaturing. Defeaturing is typically carried out manually as it requires specialist knowledge.The aim of this project is to investigate whether defeaturing can be carried out automatically using computer algorithms. The objectives are to (i) investigate methods used to represent and store computer models, their simulation parameters, and their features; (ii) develop algorithms for automated decision-making for defeaturing and implement them as new engineering software and (iii) evaluate the tools developed using selected case studies. The student will develop a computing environment for experimenting with appropriate computer algorithms. This will involve implementing simulation workflows to support the selected case studies. The student will then investigate feature and simulation parameter representation, defining a feature taxonomy that can inform the development of algorithms for defeaturing. The work will involve linking geometric features to simulation contexts. A decision-making tool will be developed that will apply defeaturing algorithms to geometries according to simulation context. A range of computer algorithms will be tested and compared. An assessment of their perceived effectiveness will be obtained by engaging with experts in industry. In summary, the project aims to deliver a context-aware defeaturing system that will improve productivity in engineering design.EPSRC research areas:- Engineering design- Geometry and topology- Graphics and visualization