Brought to you by EarthWeb
IT Library Logo

Click Here!
Click Here!

Search the site:
 
EXPERT SEARCH -----
Programming Languages
Databases
Security
Web Services
Network Services
Middleware
Components
Operating Systems
User Interfaces
Groupware & Collaboration
Content Management
Productivity Applications
Hardware
Fun & Games

EarthWeb Direct EarthWeb Direct Fatbrain Auctions Support Source Answers

EarthWeb sites
Crossnodes
Datamation
Developer.com
DICE
EarthWeb.com
EarthWeb Direct
ERP Hub
Gamelan
GoCertify.com
HTMLGoodies
Intranet Journal
IT Knowledge
IT Library
JavaGoodies
JARS
JavaScripts.com
open source IT
RoadCoders
Y2K Info

Previous Table of Contents Next


4.2. REAL-TIME ISSUES

In manufacturing fields such as robotic welding and computer control machining, recent research interests mainly tend to develop real-time expert systems for adaptive parameter adjustment. In the higher levels of plant automation, knowledge-based problem-solving techniques have been applied in domains with static data. No time-critical responses were required. In a real-time control environment, it is important to meet the strict requirements concerning the response time. It is necessary to have much smaller cycle times for real-time performance when implementing the expert system in multiple feedback control loops to munufacturing equipment. Several hardware options are implementing part of the system on a microchip, using add-on boards for signal processing in hardware and array processing for the numerical computation (Agogino, Srinivas, and Schneider, 1988).

For online control, various knowledge-based systems are embedded in a real-time environment, and therefore special attention must be paid to the time handling. Semantic networks and production rules become unmanageable as the size of the knowledge base increases, which is a very serious problem for the performance of real-time control. So, a good allocation of tasks between the expert system and the real-time control environment is necessary.

For efficient diagnosis of complex system failures in real-time environments, multiple diagnostic expert systems that cooperate with one another to solve problems beyond an individual domain of expertise are being implemented on a hierarchically distributed and cooperative processing architecture (Schwuttke, Veregge, and Quan, 1996).

A real-time control system must be composed of asynchronous concurrent processes, capable of being interrupted to accept input from unscheduled or asynchronous events. A blackboard architecture can implement a system with a multiple number of expert system kernels running in parallel, each of which can be used to solve a subproblem. Several knowledge sources can access the blackboard, like a short-term memory, and share information through the blackboard control mechanism. In progressive reasoning, the problem-solving procedure is split up into more detailed procedures, which are evaluated according to the time available. Advanced techniques for real-time requirements, such as temporal reasoning and switching of attention in reasoning, are also applicable to intelligent process control and manufacturing systems.

4.3. ADVANCED KNOWLEDGE REPRESENTATION WITH LEARNING ABILITIES

The limitation of current expert system technology is that expert systems can only use deductive inference and not inductive inference; therefore, a known starting point must be present in order to make a conclusion. In addition, expert systems have the ability to represent a human's factual knowledge, but not conceptual knowledge, and neural networks have the ability to do empirical learning, but not analytical learning. In these respects, the following target issues are important to the realization of intelligent manufacturing and engineering systems:

  1. Integration of an operational knowledge base, such as an influence diagram and a Petri net, with the neural network based learning scheme
  2. Automatic generation of the knowledge base by inductive learning from a given set of training sample operations

5. FUTURE TRENDS AND SUMMARY

Expert systems and other intelligent techniques have been widely applied to many fields in manufacturing and engineering. The current review of expert system applications has revealed the substantial effectiveness of the approaches. Developed learning and synthesizing schemes using neural networks and fuzzy logic are fairly general and could be applied to other advanced manufacturing processes such as laser or electron beam materials processing and microfabrication. In the future, unsupervised online learning and self-organizing techniques (Burke, 1992) are crucial to the development of unattended manufacturing systems to emulate a human expert fully, which obviate the need for detailed controller design or knowledge base development by hard-to-find experts and knowledge engineers.

Three difficulties, namely, hierarchical integration of various intelligent strategies, real-time performance, and knowledge representation with inductive learning abilities, have prevented conventional expert systems from producing the most efficient solutions for some problems in manufacturing and engineering. Furthermore, global optimization techniques including evaluation criteria are also practically important in optimum design and learning.

The vigorous theoretical development of artificial intelligence and control theory over the past 10 years has thus far only had a modest impact on the practice of unattended manufacturing processes, because of real-time issues and expensiveness in hardware and software realization. However, owing to the recent advances in this field and other related information technologies such as networked distributed architecures, multimedia, and human interfaces, many more expert systems that enable domain experts to develop their knowledge bases easily should be developed in order to overcome many of the weakness of conventional expert systems.


Previous Table of Contents Next

footer nav
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.