Machine understanding of three-dimensional objects for manufacturability evaluation
Abstract
A formalism for symbolic representation of a three-dimensional model is presented to aid local and global shape analysis of casting design. A three-dimensional model based on this formalism has been developed to provide meaningful geometric and topological properties for castability evaluation while providing a wide geometric coverage. Local shape analysis is carried out based on symbolic representation and global shape analysis is performed by analyzing the extracted skeleton from a discretized object. The purpose of local shape analysis is to reason about local shape characteristics so as to alter the design to meet casting requirements. The modulus approach is used to evaluate global casting soundness so as to avoid shrinkage porosity or to decide feeding path for a riser. The modulus calculation is based on the subdivision of a complex part into simple basic components and then the ratio modulus is calculated for each component and compared with the sectional modulus of direct neighbor components. The subdivision is carried out by classifying and identifying the junction types of each element of the extracted skeleton. A 3D parallel thinning algorithm has been developed in order to extract global topology of the object. This framework enables the system to reason about geometric shape based on pattern primitives and skeletons without prior knowledge about object shape. The main aim in developing the expert system is to provide the casting designer with a tool for on-line manufacturability evaluation of a part design while the functional design is being performed. With this proposed system, a smooth transition from the design to the evaluation can be expected which will lead to automated castability evaluation at the design stage so as to minimize the repeated trial and error in the casting process. An implementation of the algorithms, and examples, are provided.
Degree
Ph.D.
Advisors
Chu, Purdue University.
Subject Area
Electrical engineering|Artificial intelligence|Industrial engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.