ORCA: Outage Root Cause Analysis in DER-Rich Power Distribution System Using Data Fusion, Hierarchical Clustering and FP-Growth Rule Mining

Abstract

Power outages have been an inevitable part of power delivery systems. In case of large scale outages with conflicting information in the distribution network with distributed energy resources (DER), recognition of the root causes becomes increasingly complex. Root cause detection is very important for routing the crew and for faster restoration. Number of studies have been conducted on analysis of the outages, planning predictive actions, and improving the restoration process for minimizing the load outages using various statistical and data processing. A comprehensive methodology with the ability to find root causes in quick and efficient manner is still missing. This work proposes an Outage Root Cause Analysis (ORCA) tool, with various internal modules. The proposed ORCA utilizes the available data from diverse set of sensors (Distribution Phasor Measurement Units, metering and relay) within the distribution system (DS) to: a) aggregate using data fusion algorithm: an Ensemble Extended Kalman Filter (EEKF) approach, b) classify events at broader level (fault, breaker operation or cyber-attack with/without load outage) using hierarchical agglomerative clustering in online and offline manners, and c) narrow down the outage root cause to the possible nuanced identified causes (vegetation, animal, weather, wildfire, protection, equipment failure, planned) using the available historical data and the Frequent Pattern-Growth data mining approach. Simulation results demonstrate the superiority of the proposed approach compared to the other existing key approaches using two different test systems and multiple outage with different locations case scenarios.