Distribution Grid Fault Diagnostic Employing Hilbert-Huang Transform and Neural Networks

Abstract

Faults in distribution grids cause power interruption and economic losses. A crucial part of distribution grids protection systems is effectively diagnosing the fault to accelerate the power restoration process. This paper presents a fault diagnostic method for a distribution grid that consists of the Hilbert-Huang transform (HHT) and feedforward neural networks (FFNN). First, instantaneous amplitude (IA) and frequency (IF) are obtained from the HHT. Subsequently, statistical features are extracted from IA and IF plots and fetched to the FFNN for detection, classification, and location identification of different types of faults. The proposed approach is tested on a distribution grid modeled in MATLAB/SIMULINK platform. Obtained results demonstrate the effectiveness of the developed method for both noise-free and noisy data with the variation of pre-fault loading conditions, fault resistance, location, and inception angle.