![]() |
|||
![]()
|
![]() |
![]() |
![]() |
Part III
|
1. | Diagnosis | |
2. | Diagnostic Techniques and Methods | |
3. | Diagnostics Conceptualization | |
3.1. | What is Conceptualization? | |
3.2. | Components of Conceptualization | |
3.3. | Identification of Strategic, Tactical, and Factual Knowledge | |
3.4. | Construction of the Static Submodel: Concept Dictionary | |
3.5. | Creation of the Dynamic Submodel | |
3.6. | Knowledge Map Generation | |
3.7. | Verifications | |
3.8. | System Decomposition | |
4. | Conclusions | |
References |
Experts in a domain execute generic tasks that define how competent they are in their specialty. These tasks are cataloged either hierarchically, as shown in Figure 1 (Schreiber, 1993) or in a table (Maté, 1989; Liebowitz, 1989; Hayes-Roth, 1983).
FIGURE 1 Hierarchical classification of known problems in knowledge engineering
FIGURE 2 Percent application of expert systems by problem type.
An examination of these tasks provides an understanding of what makes expert problem-solving behavior difficult. Of these generic tasks, which identify a host of technical solutions to particular problems, the most universally applied is, as shown in Figure 2, unquestionably what is referred to as "diagnosis."
As if the mere extension of the diagnostic task were not enough to underline its importance, consider the possible consequences of not knowing the exact and precise cause of a malfunction in a nuclear power plant. The resulting disasters, Chernobyl is a paradigmatic example of this, are costly in terms of material (cost of the plant, patient treatment, etc.), human (immediate and later deaths from radioactivity), and ecological (destruction of the environment) losses. The above can be extended to domains as far apart as the explosion of Challenger in 1986, when Judith Resnik, the first American woman to go into space, died, together with her fellow astronauts, and its financial cost; or faults in aviation systems, causing planes to crash with a dismal sequel of deaths, etc.
Therefore, whatever the system, it is vitally important to be able to accurately determine the faulty subsystem or component. There are two main reasons for this. First, to take precautions against and, as far as possible, prevent disasters and accidents actually occurring. Second, today's systems are becoming more and more sophisticated and, taking into account the potentially devastating consequences of a system fault, correct and meticulous maintenance is a vital and extremely difficult exercise, making diagnosis a demanding task, relying heavily on expertise, as a large amount of knowledge has to be assimilated and rapidly and expediently accessed.
Although this maintenance and control work is carried out by expert operators, this is often insufficient, as human capabilities are overwhelmed by the requirements of these systems. Their operation usually requires enormous computing power and real-time performance, such as short response time and continual adaptation to changes in system variables.
We can conclude that human beings are incapable of efficiently controlling systems, such as those described above. Fortunately, human beings' limitations in the field of industrial and natural diagnosis can be overcome by developing computer technology. Computers can react faster than human beings, filter the information received, and show only the most important information or the information most relevant in a given situation. It is in these circumstances that a knowledge-based system, and particularly an expert system (ES), can be of great use, because it has the capability of rapidly and efficiently processing huge amounts of knowledge. Therefore, there is a tendency today to perform diagnostics with the aid of computer systems that enable the operator to perform his/her work safely, comfortably, rapidly, and reliably. Expert systems began to be used in diagnostics because human beings were found to be poor diagnosticians. The reasons (Krishna, 1991) for this are operator inexperience in complex environments; the stress to which the operator is subject because of the need to make quick decisions about systems, when a slow or mistaken action could lead to irreparable damage; slowness of reaction, which usually occurs precisely in events that require utmost expediency; information overload (human beings are only capable of assimilating a limited amount of input information and they can be clearly overwhelmed by the enormous amount of data produced by the system under control); limited processing capability (our calculation capability is insufficient to perform all the calculations required), and fatigue (human beings naturally get tired and bored, which leads to errors).
Though there are as many definitions of diagnosis as there are authors who address the question, all have a common denominator. Diagnosis, from the Greek [delta][iota][alpha], through, and [gamma][nyo][omega][sigma][iota][zeta], knowledge, means the analysis of system malfunction from noisy information or observable or non-observable facts related to both its behavior or function and its structure. The information may be quantitative, and therefore objectifiable and measurable, what are referred to as signs, or qualitative, "detectable" or subjective, called symptoms. Faults may sometimes be masked by signs and/or symptoms of other faults or they may be intermittent, in which case a system has to be forced to reveal faults. The diagnostic equipment itself may even fail. This category of problems includes electronic, financial, industrial, medical, social, software diagnostics, etc.
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. |