Off-campus UMass Amherst users: To download dissertations, please use the following link to log into our proxy server with your UMass Amherst user name and password.
Non-UMass Amherst users, please click the view more button below to purchase a copy of this dissertation from Proquest.
(Some titles may also be available free of charge in our Open Access Dissertation Collection, so please check there first.)
A manufacturing based approach to preliminary design using coupled symbolic and numerical computing techniques
Abstract
This research advocates a unique design-by-manufacturability approach for providing deep science-based knowledge to feature based design systems. The method provides yet another viable tool for concurrent engineering. Unlike the shallow rule based systems, the proposed system extracts and exploits deep physical knowledge of the manufacturing process, thereby, rendering greater intelligence to design tools. The methodology needs to abstract deep knowledge from numerical simulation results of the design's manufacturing process into the design stage. However, due to the lack of tight coupling between design and analysis algorithms, suitable representations are necessary to link design and analysis modules within a single computational framework. The nature of this research dictates the need for coupling symbolic and numerical computing techniques for formulating a representative language. Representational adequacy is necessary for providing a front-end link to the analyzer, while inferential adequacy is essential in bringing the deep knowledge back to the solid modeler. Suitable representations have therefore been formulated to link the analyzer with the solid modeler. Secondary characteristics were derived from the primary representation of the design for activating the numerical solver. The raw numerical data from the solver is the source of deep physical knowledge for the computational system. Symbolic manipulations were performed to transform this low level numerical information to high level abstractions suitable for feature-based design modification. A "blackboard" type architecture was used for organizing the solution space with information obtained at different levels of abstraction. The data is arranged along the vertical dimension which separates and distinguishes entries at different levels of the design process. Thus, the numerical information is available at the lowest level of abstraction, while a qualitative description of the design exists at the top most level. Each entry in the blackboard activates a different knowledge source for information updates. A proof-of-concept program was developed to demonstrate these ideas. The system capabilities have been demonstrated in a X-Windows environment using the solid modeling and analysis modules of I$\sb-$DEAS*. The system was tested for the design of thin shelled plastic components. In this case, deep knowledge of the injection molding process was abstracted from moldfilling simulations to develop the interpretation expert. Examples are presented illustrating these concepts and validate the design by manufacturability approach. ftn*I$\sb-$DEAS is an integrated design and analysis software package developed and distributed by Structural Dynamics Research Corporation
Subject Area
Mechanical engineering|Artificial intelligence
Recommended Citation
Sahu, Kaushik, "A manufacturing based approach to preliminary design using coupled symbolic and numerical computing techniques" (1992). Doctoral Dissertations Available from Proquest. AAI9233156.
https://scholarworks.umass.edu/dissertations/AAI9233156