Wavelet and ANN Based Relaying for Power Transformer Protection

Abstract

This paper presents an efficient wavelet and neural network (WNN) based algorithm for distinguishing magnetizing inrush currents from internal fault currents in three phase power transformers. The wavelet transform is applied first to decompose the current signals of the power transformer into a series of detailed wavelet components. The values of the detailed coefficients obtained can accurately discriminate between an internal fault and magnetizing inrush currents in power transformers. The detailed coefficients are further used to train an Artificial Neural Network (ANN). The trained ANN clearly distinguishes an internal fault current from magnetizing inrush current. A typical 750 MVA, 27/420KV, /Y power transformer connected between a 27KV source at the sending end and a 420KV transmission line connected to an infinite bus power system at the receiving end were simulated using PSCAD/EMTDC software. The generated data were used by the MATLAB software to test the performance of the proposed technique. The simulation results obtained show that the new algorithm is more reliable and accurate. It provides a high operating sensitivity for internal faults and remains stable for inrush currents of the power transformers.

A combinatorial approach based on wavelet transform and Hidden Markov Models in differential relaying of power transformers

“…To avoid the tripping operation dependency to such a threshold many novel methods have been proposed. These approaches includes: autoregressive process based on power spectrum of harmonics [2], wavelet transform based schemes [3][4][5][6][7][8][9], artificial neural networks [8][9][10], leakage inductance based techniques [11,12], harmonic characteristics [13], induced voltage based method [14], correlation transform [15] and Hidden Markov Model (HMM) [16]. The main drawbacks of priori methods are determining an exact restrain threshold, long data window requirement, sensitivity to noise, delayed fault detection and voltage transformers requirements.…”

A fast WT‐based algorithm to distinguish between transformer internal faults and inrush currents

“…To discriminate between internal faults and inrush currents other approaches such as power differential method, transformer inductance variation, flux and voltage restraints and artificial neural networks are presented [6][7][8][9][10]. These approaches depend on parameters of the protected transformer.…”

A new stochastic method based on Hidden Markov Models to transformer differential protection

“…To avoid the tripping operation dependency to such a threshold many novel methods have been proposed. These approaches includes: autoregressive process based on power spectrum of harmonics [2], wavelet transform based schemes [3][4][5][6][7][8][9], artificial neural networks [8][9][10], leakage inductance based techniques [11,12], harmonic characteristics [13], induced voltage based method [14] and correlation transform [15]. The main drawbacks of priori methods are determining an exact restrain threshold, long data window requirement, sensitivity to noise, delayed fault detection and voltage transformers requirements.…”