Fault Location in Power Distribution Networks using Arbitrary Similarity Criteria in the Principal Component Subspace

This paper presents a new strategy to support fault location in power distribution networks based on arbitrary similarity criteria in the principal component subspace. Input data consist of impedance measurements at the secondary transformer of one or more substation buses, which are used to build data-driven models of the grid operating conditions. The statistical models of actual operating conditions are further compared with a few reference scenarios to determine the network configuration and the type and location of the fault based on arbitrary features which minimize the variability of the data. Furthermore, this paper includes a case study with a real-based low voltage power distribution network to test the method under different faults.