Ground Fault Identification and Key Feature Extraction Method for Distribution Network Based on Waveform Analysis

Abstract

Because the fault current is weak and difficult to be identified, a method for ground fault identification and key feature extraction in distribution network based on waveform analysis is proposed. By analyzing the mutation characteristics and transient characteristics, waveform analysis is used as the feature extraction method, combined with the normalization processing method, to obtain the target feature components. Identify fault persistence features, extract frequency band components, and obtain a set of pulse signals through mathematical morphological transformation. The positive impulse noise and negative impulse noise fault signals extracted are suppressed by combining the opening operation and the closing operation. After analyzing the characteristic quantity of distribution network, the characteristic parameters of fault identification are determined. The volt-ampere characteristics of linear distribution network components are analyzed, and fault line identification is realized according to the characteristic components. Analyze metallic ground fault, arc ground fault, and intermittent arc ground fault waveforms, divide characteristic areas, and complete ground fault identification. The experimental results show that the current transient component fluctuation curve of this method is consistent with the actual fluctuation curve, and the maximum identification accuracy and identification time are 0.988 and 20 s respectively, experiments show that this method has high accuracy and recognition rate.