Presidential Young Investigator Award: Feature-based Manufacturing

这位总统青年调查员的研究有两个目标。第一个目标是创建软件技术，根据存储在计算机辅助设计（CAD）系统中的部件描述作为实体模型来自动估计制造成本。制造成本是加工过程的函数，反过来，加工过程是零件特征的函数。所提出的成本估计器由两个模块组成：特征识别器和特征解释器。识别器作用于实体模型边界表示来提取形状特征。解释器在选择制造工艺和确定加工成本时使用这些特征。除了能够在设计过程中估算成本的明显好处之外，估算器还将帮助确定用于制造的一组有用的表单特性。第二个目标是开发一种特征语言，可用于根据几何和拓扑原语（如面、边、顶点和顶点-边缘循环）描述特征。特性在CAD中的使用还处于起步阶段，主要是因为还没有开发出完整的分类法。分类法依赖于基于某些特征对功能进行排序的能力。具有良好语法的功能语言将允许为各种类型的功能编写明确的定义。候选语言结构包括句法模式语法和图语法。这项研究的结果将产生一个特征概要，以及一个有组织的数学方法来定义和存储它们。这一进步将有助于那些涉及特征识别和基于特征的设计的人。

The research of this Presidential Young Investigator has two goals. The first goal is to create software techniques to estimate automatically manufacturing costs based on a part description stored in a computer-aided design (CAD) system as a solid model. Manufacturing costs are a function of the machining process, and, in turn, the machining process is a function of the features of a part. The proposed cost estimator consists of two modules: a feature recognizer and a feature interpreter. The recognizer acts on a solid model boundary representation to extract form features. The interpreter uses the features in selecting manufacturing processes and determining processing costs. In addition to the clear benefit of being able to estimate costs during the design process, the estimator will help determine a useful set of form features for manufacturing in general. The second goal is to develop a feature language that can be used to describe features in terms of geometric and topological primitives such as faces, edges, vertices, and vertex-edge loops. The use of feature in CAD is in its infancy, primarily because no complete taxonomy has been developed. The taxonomy depends on the capability to sort features based upon certain characteristics. A feature language with a well-formed grammar would allow unambiguous definitions to be written for various classes of features. Candidate language constructs include syntactic pattern grammars and graph grammars. The results of this research will yield a compendium of features together with an organized and mathematical way to define and store them. This advances will aid those involved in both feature recognition and feature-based design.