Internet based equipment maintenance system using Bayesian network.
Date of Issue2000
School of Mechanical and Aerospace Engineering
In this thesis a web based equipment maintenance system has been developed to help the equipment supplier in the event of equipment failure. The main part of the system is the Java program developed upon the Bayesian network theory and having problem solving ability of an expert. The Java program operates in two parts. In the first part the user carries out a few tests, specified by an expert and the results obtained are input into the graphic user interface (GUI) of the system. In the second part the GUI transfers the results as input to the Bayesian network which computes the probability of failure of each machine component. The component having highest probability of failure is checked first followed by other components in the descending order. This system can also generate results under uncertainty. In the actual production line, due to constraint, it may not be possible to carry out all the tests. As later shown in this thesis, even in such cases, the Bayesian network is able to generate results with acceptable accuracy.
Nanyang Technological University