An expert system methodology for determining the characteristics of power system component failures

Abstract

The objective of the paper is to describe an expert system methodology which has been developed for determining the most likely nature, cause and expected duration of failures occurring in power system components. The inference engine task is based on Bayes' theorem and the knowledge acquired during the study of the historical outage records of the power system. This study allows the expected values of outage rates and durations to be evaluated for all component categories and groups. This expert system methodology can be a very useful tool with which to determine the most likely expected duration of a failure and to select the most appropriate operational actions to be performed. An example of application of this methodology to failures occurring in power transformers is also included.