An Effective and Computationally Efficient Similarity-Based Method for Fault Region Identification in Electrical Distribution Systems

ABSTRACT: Nowadays, fast fault location in distribution systems plays a vital role for both utilities and end-users, improving grid availability and thereby reducing utility penalties and minimizing disturbances in consumers’ daily lives. Therefore, this work presents a simple yet accurate and robust method for identifying the faulted region in electrical distribution systems. To perform this task, the proposed method requires only local fault current measurements (available from digital relays) and a database built from historical data, simulations, or both, thereby eliminating the need for additional hardware or communication links. The Hausdorff distance (HD) was used to measure the similarity level between a fault current waveform acquired from the field and reference current waveforms previously stored in a database, thereby implementing a waveform pattern-matching approach. The method was initially developed and tested on a small distribution system, and subsequently validated using the IEEE 34-bus test feeder. After a sensitivity analysis regarding the most suitable data window length, sampling frequency, and input signals, the results demonstrate that the proposed method is reliable and accurate, even under different fault characteristics and loading conditions.