Data-Driven Fault Localization in Distribution Systems with Distributed Energy Resources

The integration of Distributed Energy Resources (DERs) introduces non-conventional two-way power flows which cannot be captured well by traditional model-based techniques. This brings great challenges to accurately localize faults and initiate correct actions of the protection system. In this paper, we propose a data-driven fault localization strategy based on multilevel system regionalization and probabilistic fault detections on all the subregions. The strategy combines the Support Vector Data Description (SVDD) and the Kernel Density Estimation (KDE) to provide the confidence level of fault detections in each subregion by p-values, and then accurately localize the fault by comparing the p-values. Our experiments show that the proposed data-driven fault localization can greatly increase the accuracy of fault localization for distribution systems with high integration of DERs.