Research on Distribution Network Fault Recognition Method Based on Time-Frequency Characteristics of Fault Waveforms

Automated Fault Analysis and Diagnosis using High-Frequency and Maintenance Data from Distribution Networks

“…If a fault has already occurred, the root cause of the fault can be identified through combining the context with the waveform characteristic of a new event. The features selected are referenced to the ones in [10][11] which uses weather observations from public stations and pertinent waveform characteristics. For demonstration purposes, one of the simplest classifiers, K-Nearest Neighbors (KNN), is used here to identify fault causes from the waveform shapes; benchmarking against a number of state of the art classifiers revealed this to be the superior performer although a formal model selection is left to future work, with this model pursued for completeness of demonstration purposes.…”

A high frequency reflected current signals-based fault type identification method

“…Some fault type classification methods used the artificial intelligent methods, such as support vector machines [9][10][11], artificial neural network (ANN) [12], ANN with the use of particle swarm optimization (PSO) [13] and feedforward neural network combined with S-transform [14]. The fault classification algorithms identify various fault types based on sample values of voltage and current signals compared with their predefined values in [15] and time frequency characteristics of fault waveforms in [16]. Moreover, to enhance performance of directional relaying, fuzzy inference system is used to detect fault type with the help of magnitude of fundamental current and voltage signals [17].…”

Fault‐cause identification method based on adaptive deep belief network and time–frequency characteristics of travelling wave

“…The principle of the method is to judge the fault line according to the characteristics of the electrical quantity during the fault time, and then the fault cause can be identified based on the meteorological conditions and manual inspection. The common fault root causes can be categorised as branches, mountain fire, animals, lightning, icing, wind, and external damage [2][3][4].…”

“…Furthermore, existing machine learning methods mainly classify faults from the perspective of line selection, and few studies have focused on a method to directly classify and identify the fault root cause. Moreover, the existing fault-cause identification method has certain limitations in identifying the number of fault types or the recognition accuracy [2][3][4].…”