A Method of Failure Causes Analysis Based on Combination of Metering Abnormal Events*

The existing abnormal electrical behavior analysis is based on single metering abnormal event. It has limitations such as heavy workload and low accuracy. In order to improve the utilization of abnormal data and the accuracy of abnormity diagnosis, this paper proposes a method to analyze the fault causes based on the combination of abnormal events. Firstly, a correlation degree calculation method is determined based on the intervals of failure occurrence date. Next, thresholds are set to identify concurrent abnormal events (CAE). Finally, a method to analyze the fault causes has been proposed by combining the correlation degree with the probability of the fault causes. The results indicate that this method can improve the accuracy of fault diagnosis.