High impedance fault detection in distribution networks using randomness of zero-sequence current signal: A detrended fluctuation analysis approach

Detection of high impedance faults (HIFs) in low and medium voltage distribution networks has always been challenging due to their lower magnitude and random current characteristics. Conventional overcurrent relays are ineffective in detecting HIFs because of the low fault current associated with them. Consequently, existing HIF detection schemes predominantly rely on features extracted from the frequency, time–frequency, or symmetrical component domains. However, these methods are often limited in their effectiveness under specific conditions, such as particular voltage levels, conductor types, or environmental factors, due to the multifaceted nature of HIF currents, which depend upon a variety of factors such as type of materials, voltage level, wetness of the surface, shape of the conductor, and even the weather conditions. On the other hand, due to the intrinsic presence of arcing in the HIF phenomenon, the resultant fault current consistently assumes a random character, and this inherent randomness can be leveraged as a potential feature for fault detection. This paper proposes a new HIF detection method by analyzing the randomness or unpredictability of the fault current signals. The Detrended fluctuation analysis (DFA) is innovatively employed to assess the unpredictability of the lower frequency component of the zero-sequence current by examining its instantaneous amplitude envelope. The accuracy of the proposed approach is verified with extensive simulation data of fault events under diverse operative conditions. The security of the proposed scheme is also verified under several no-fault transient circumstances, such as capacitor switching and load perturbation. The computational efficiency of the method has been verified through the process-in-loop (PIL) simulation using cost-effective hardware.