![]() |
|||
![]()
|
![]() |
![]() |
![]() |
Chapter 22
|
1. | Introduction | |
2. | Historical Background | |
3. | Methodologies, Techniques, and Practices | |
3.1. | Intelligent Process Control | |
3.2. | Knowledge-Based Adaptive Control in Machining | |
3.3. | Machine Learning Using Neural Networks | |
3.4. | Genetic Algorithms in Manufacturing and Engineering | |
3.5. | Process Diagnosis Using Probabilistic Inference | |
3.6. | Failure Diagnosis Using Case-Based Reasoning | |
3.7. | Expert Systems for Welding | |
3.8. | Neural and Fuzzy Rule-Based Expert Systems for Automated Welding | |
3.9. | Expert System Development Tools | |
3.10. | Knowledge Representation Using High-Level Petri Nets | |
4. | Research Issues | |
4.1. | Hierarchical Integration of Intelligent Strategies | |
4.2. | Real-Time Issues | |
4.3. | Advanced Knowledge Representation with Learning Abilities | |
5. | Future Trends and Summary | |
References |
The purpose of this chapter is to illustrate and discuss some uses of expert system techniques in manufacturing and engineering fields. The fields include integrated computer-aided manufacturing with real-time adaptive control in the metal fabrication industry, where machining and welding are the two major categories, although many expert systems have also been developed in such manufacturing fields as grinding, injection molding, electrical discharge machining, assembly, etc. Owing to rapid progress in microelectronics, computer science, and control theory over the past 30 years, there has been a very significant development of automation techniques aiming at unattended manufacturing. Today's industrial robots have efficient and reliable abilities for simple material handling or welding tasks in regulated and fixed environments of factories. In engineering fields, integration of CAD with artificial intelligence techniques has surprisingly progressed to develop design-with-feature systems and image understanding systems based on human mental processes.
However, it is observed that the actual implementation of computer control systems for manufacturing control often incorporates a substantial amount of human operators' knowledge or skills and heuristic control logic. This is true for simple feedback control laws as well as for more sophisticated multivariable control loops with many interacting process variables, as required in most of manufacturing processes. Consequently, well-functioning control systems should be provided with a more considerable amount of heuristic logic for selection and supervision of operating conditions carried out by lower level controls. Human operators' knowledge and skills concerning equipment operations are often more efficient than present automated systems, but not easily encoded by numerically based algorithms for optimization and control. Thus, development of intelligent control schemes that can learn and synthesize knowledge effectively in automated manufacturing is urgently required because of the serious shortage of experienced operators in industry.
Basically, the expert system paradigm enables the separation of the inference mechanism and the domain knowledge. The inference mechanism is domain independent. This separation allows the user to deal easily with the knowledge base and the inference structure, and provides a systematic approach for applying heuristic logic using human operators' knowledge or skills in manufacturing and engineering.
In this chapter, first, the historical background of the needs of expert systems in manufacturing and engineering is discussed with the nature of multilayered manufacturing processes. Welding is taken as a representative example because it is considered to be one of the most difficult fields for automation. Methodologies, techniques, and practices in existing expert system applications, especially intelligent process control, adaptive machining control, selection of optimum machining parameters, process monitoring, and failure diagnosis, are critically reviewed. Then, an expert system that integrates welding procedure selection and welding process control is briefly described as a whole. Finally, research issues with respect to methodologies, especially concerning the implementation of integrated and real-time expert systems, and advanced knowledge representation with inductive learning abilities, are briefly discussed.
Previous | Table of Contents | Next |
![]() |
|
Use of this site is subject certain Terms & Conditions. Copyright (c) 1996-1999 EarthWeb, Inc.. All rights reserved. Reproduction in whole or in part in any form or medium without express written permission of EarthWeb is prohibited. Please read our privacy policy for details. |