Abstract
This dissertation describes a new Concurrent Engineering system with distributed intelligence to demonstrate the use of a Virtual Round Table to assist a designer in the early design stages of a spot welding application. This Computer Aided Process Planning system is called a Multi Expert System. The first system consisted of four expert systems: Computer Aided Design Expert, Material Expert, Spot Welding Expert and Production Expert.A limitation of existing systems was that they were designed to model the product and process design as a whole, which made the system inflexible for expansion when new processes needed to be included.
The new system described in this dissertation overcame this problem by modelling product and process design using a natural approach for capturing intelligence from different processes. The new approach categorised the product and process design processes into separate individual experts. Intelligence and knowledge from each process was captured and embedded within the individual expert that represented the process. A virtual round table for design was created. This approach enabled greater flexibility in terms of modifiability, upgradeability, extendibility and reusability.
To demonstrate this new approach, a new CAD Expert was created. The expert was used to interpret drawing files from CAD software. A new simple and efficient method to detect overlaps that used an entities comparison technique is presented. The material selection process was modelled with a Material Expert. A new property weighting method was formulated and successfully tested to improve the materials selected.
A Spot Welding Expert was created to help designers in spot welding tasks. A new method for spot welding knowledge acquisition and representation is presented. The new method used a novel "Expandable Knowledge Base" that allowed knowledge to be expanded over time and experience. In addition, this technique enabled a suggestion or an estimated answer to be calculated from existing knowledge even if the specific knowledge required was not available. Spot welding locations were then determined.
The manufacturing process was integrated with the new system as a Production Expert. Fanuc S-700 robot programs were automatically generated for spot welding tasks. The time and cost needed to perform a task was calculated and the programs were used to drive a Fanuc S-700 robot.
Finally, to prove the flexibility of the system, a new virtual round table for a spray painting application was created. This work was an extension of research by Sanders(1994) and Tewkesbury (1994) on design using distributed intelligence within Advanced Production Machinery and Rasol (2000) on automatic systems for simple spot welding tasks.
| Date of Award | 2007 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | David Sanders (Supervisor) |