High-speed directional protection without voltage sensors for distribution feeders with distributed generation integration based on the correlation of signals and machine learning

Abstract

This paper proposes a novel methodology to define fault current direction along the Distribution Feeder (DF) considering Distributed Generation (DG) integration. The proposed methodology is based on Empirical Decomposition (ED), Decision Trees (DT) and Support Vector Machine (SVM). Using ED, it is possible to determine different Principal Components (PCs) that are used are inputs in these DT and SVP classifiers. Assessment of methodology considering different faults, inception angles, fault distances, and others are carried out. Besides, the proposed methodology is tested successfully considering different distribution system topologies and by analyzing special features required by relay manufacturers. Test results highlight the efficiency of the methodology, which presents a concise design and a simple mathematical formulation in the time domain.