An Approach to Detection of High Impedance Fault Using Discrete Wavelet Transform and Artificial Neural Networks

Abstract

High impedance faults (HIF) are faults that are difficult to detect by conventional protection relays. In this paper, a new HIF model is introduced and a novel methodology is presented to detect HIF by means of discrete wavelet transform (DWT) and artificial neural network (ANN). The distorted waveforms (HIF, load switching, line switching, capacitor switching and non-linear loads that behave similar to HIF current) are generated using PSCAD/EMTDC, captured with a sampling rate of 20 kHz and de-noised using DWT to obtain signals with higher signal-to-noise ratio. DWT is used to decompose the distorted signal and to extract its useful information. Appropriate feature vectors are created and applied in training the ANN. The effectiveness of the proposed method was tested using a wide spectrum of disturbances. Simulations are carried out to confirm the suitability and capability of the proposed method in HIF detection.

Detection and Location of High Impedance Faults in Distribution Systems: A Review

“…Fault separation was related to a most extreme mistake level of 0.688%. In this way, the outcomes affirm the dependability and reasonableness of the proposed technique under various fault circumstances [11,12].…”

“…ANN [8]- [12] Effortlessness in usage. Recognizes the nonlinear connection amongst needy and free factors.…”

Hybrid detection algorithm for online faulty sensors identification in wireless sensor networks

“…Mother wavelet selection is an important part of WT, the choice depends on the target applications [49]. There are many types of wavelets such as Haar, Daubechies (db1, db2,…), Symlets (sym1, sym2,…) and Coiflets.…”

High impedance fault detection and identification based on pattern recognition of phase displacement computation

“…[6][7][8][9][10], Fourier transform (FT) is utilized to extract fundamental, even harmonic, odd harmonic, and inter harmonic frequency components of the voltage during an HIF event. Besides FT, wavelet transform (WT) has been widely used because of its capability to provide both time and frequency localization [11][12][13][14][15][16][17][18][19][20]. Usually, WT is used to extract important features from the signal.…”

“…Then, a classifier or pattern recognition method is utilized to classify the event based on the features. These pattern recognition methods include artificial neural network (ANN) [13,14], fuzzy system (FS) [16,21], nearest neighbor rule (NNR) [17], Bayes classifier [18], moving window approach [12] and deterministic logic scheme [15].…”