A neural network hybrid expert system
- Portland State Univ., OR (USA). Dept. of Mechanical Engineering
When knowledge-based expert rules, equations, and proprietary languages extend Computer Aided Design and Computer Aided Manufacturing (CAD CAM) software, previously designed mechanisms can be scaled to satisfy new design requirements in the shortest time. However, embedded design alternatives needed by design engineers during the product conception and rework stages are lacking, and an operator is required who has a thorough understanding of the intended design and the how-to expertise needed to create and optimize the mechanisms. By applying neural network technology to build an expert system, a robust design supervisor system emerged which automated the embedded intellectual operations (e.g. questioning, identifying, selecting, and coordinating the design process) to (1) select the best mechanisms necessary to design a power transmission gearbox from proven solutions; (2) aid the inexperienced operator in developing complex design solutions; and (3) provide design alternatives which add back-to-the-drawing board capabilities to knowledge-based mechanical CAD/CAM software programs. 15 refs., 2 figs.
- Research Organization:
- Los Alamos National Laboratory (LANL), Los Alamos, NM (United States)
- Sponsoring Organization:
- USDOE; USDOE, Washington, DC (USA)
- DOE Contract Number:
- W-7405-ENG-36
- OSTI ID:
- 6090218
- Report Number(s):
- LA-UR-91-903; CONF-9103136-3; ON: DE91009956
- Resource Relation:
- Conference: 5. national conference on undergraduate research, Pasadena, CA (USA), 21-23 Mar 1991
- Country of Publication:
- United States
- Language:
- English
Similar Records
CAD/CAM --on the integration of CAPP/CACE for the generation of least-cost machining-process plan for multi-feature work pieces
CAD-CAM database management at Bendix Kansas City