Power distribution system fault cause analysis by using association rule mining

In recent years, with the increasing requirements on power distribution utilities to ensure system reliability and to improve customers and regulators satisfaction, utilities seek to find practical solutions that enable them to restrict specific faults or to better manage their responses to unavoidable power outages. For achieving either, it is crucial to acquire a profound understanding of different faults by exploring their underlying causes and identifying key variables related to those causes. Currently, statistical models as well as advanced data analytics techniques are common tools to gain such understanding. Although basic statistical analysis provides a general knowledge of the primary causes of faults; nevertheless, it falls short of describing nuanced conditions that lead to a fault. On the other hand, applying sophisticated algorithms can produce deeper insight into the main causes; however, it would be computationally burdensome and might require a tremendous amount of running time. In order to overcome these problems, this paper proposes a novel approach for fault cause analysis by using association rule mining. The primary goals are to characterize faults according to their underlying causes and to identify important variables that strongly impact fault frequency. This paper proposes a step-by-step procedure, which deals with data preparation, practical issues associated with fault data sets, and implementation of association rule mining. The procedure is followed by a comprehensive case study to demonstrate how the proposed approach can be used to mine for causal structures and identify frequent patterns for vegetation, animal, equipment failure, public accident, and lightning-related faults.