High-impedance fault detection method based on sparse data divergence discrimination in distribution networks

Abstract

Due to the weak fault current characteristics, strong nonlinearity and uncertainty, and presence of similar transient features of switching events, it is difficult for traditional methods to accurately detect and identify high faults in a high-noise environment. This paper proposes a high-impedance fault detection method based on sparse data divergence discrimination. First, sensitive sparse principal component analysis is applied to extract the joint features contained in the measurement data. Then, according to the sparse data and singular covariance matrices, an improved Wasserstein divergence solution method is proposed to obtain the difference between data probability distributions before and during a fault. Finally, threshold and time criteria are applied to discriminate against high impedance faults in high-noise cases. The results show that the proposed method can accurately detect and identify high-impedance faults with good robustness to the fault inception angle, transition impedance, noise, and switching events.© 2017 Elsevier Inc. All rights reserved.