Files and links (1)
PDFOpen Access (License Unspecified), Open Access
Thesis
Automatic classification of CAD models
Master of Science (M.S.), Drexel University
Apr 2005
DOI:
https://doi.org/10.17918/etd-505
Abstract
Classification of CAD helps the reuse of engineering designs and accelerates productdevelopment. Existing research, based on either group technology or fixed modelingmatching algorithms, impose a priori categorization schemas on engineering data or require significant human labeling. This research introduces a framework for automatic classification of CAD models using a pattern classification model. It separates feature extraction and classification from 3D shape matching procedures. Different shape features andclassifiers can be incorporated to fit example CAD/CAM oriented classification schemas. Two example approaches are presented to demonstrate how CAD models can be classified with supervised machine learning. The first approach describes how to perform nearest neighbor shape classification with multiple descriptors using a weighting formulation. Appropriate weights of different shape descriptors are estimated through training. The second approach aims at classifying prismatic-machined, and cast-then-machined artifacts. These manufacturing processes are discrimated by normal surface curvature and nonlinear support vector machines.
Metrics
53 File views/ downloads
5 Record Views
Details
- Title
- Automatic classification of CAD models
- Creators
- Cheuk Yiu Ip - DU
- Contributors
- William Clement Regli (Advisor) - Drexel University (1970-)
- Awarding Institution
- Drexel University
- Degree Awarded
- Master of Science (M.S.)
- Publisher
- Drexel University; Philadelphia, Pennsylvania
- Resource Type
- Thesis
- Language
- English
- Academic Unit
- College of Arts and Sciences; Drexel University; Mathematics
- Other Identifier
- 505; 991014632173904721