Journal of the Japanese Society for Artificial Intelligence
Online ISSN : 2435-8614
Print ISSN : 2188-2266
Print ISSN:0912-8085 until 2013
A Fault Location Expert System for Subscriber Network
Tetsuji ABEYukio AKIYAMAYutaka MITSUNAGA
Author information
MAGAZINE FREE ACCESS

1990 Volume 5 Issue 2 Pages 155-162

Details
Abstract

Nowadays, a huge-scale telecommunication network has been constructed in Japan, and high level maintenance technology is required to keep the system reliable. Especially, since the subscriber network has intricate structure and many outside equipments, the maintenance of it is difficult, and a expert system is expected to make the reliability higher. In this paper, a subscriber network fault location expert system developed as one of the maintenance system is described. By investigating the fault location problem and characteristics of knowledge which we can collect, it is found that the principle knowledge of this problem is past fault facts of each equipment of the system. Thus, a reasoning method that utilize knowledge of past fault facts effectively is proposed. The reasoning method infers the failure possibility of every equipments using a concept of failure physics. It is proved that fundamental failure factors must be derived from the past fault fact to avoid the overlap or contradiction of knowledge. Then, a method to obtain those factors is also proposed. It is shown that the failure factor can easily derived using this method. A prototype system is also developed. Because the reasoning method is based on the failure physics, the reasoning results have objectivity. And the system also has an advantage that the fault facts knowledge can be simply represented by the frame. Consequently, it is confirmed that the system has enough capacity for practical application.

Content from these authors
© 1990 The Japaense Society for Artificial Intelligence
Previous article Next article
feedback
Top