A method to capture and de-noise partial discharge pulses using discrete wavelet transform and ANFIS

Abstract

Summary Due to the presence of excessive noise in the recorded partial discharge (PD) current signals, de‐noising of these signals is a crucial task for performing any investigation on the subject. Meanwhile, to accelerate this de‐noising process a single PD pulse can be extracted from the train of those recorded pulses, followed by its de‐noising. In this paper a single PD pulse is extracted from the train of recorded PD pulses, using noisy recorded data cumulative energy. A de‐noising technique based on adaptive neuro‐fuzzy inference systems is proposed. To verify the validity of the proposed method, four different sources of PD signals are physically simulated. The proposed method is applied on laboratory recorded signals and is compared with a conventional method which is based on discrete wavelet transform (DWT) and linear adaptive filtering method. Furthermore, the proposed algorithm is used for extraction and de‐noising of single PD activities of an on‐site recorded data. This method introduces an efficient pulse distortion mitigation tool which requires no adjustment of any pre‐set parameter—neither for different noise levels, nor for different kind of PD pulses. This is an important feature of the proposed technique. Copyright © 2014 John Wiley & Sons, Ltd.

Partial Discharge Signal Extraction Method Based on EDSSV and Low Rank RBF Neural Network

“…At present, a variety of methods have been proposed on noise suppression of PD signals, the common methods are wavelet transform (WT) [7]- [8], empirical modal decomposition (EMD) [9]- [10], and singular value decomposition (SVD) [11]. The WT features good timefrequency analysis ability.…”

A novel discrete wavelet transform‐based graphical language classifier for identification of high‐impedance fault in distribution power system

“…In this work, the feature extraction has been carried out using Fast Fourier Transform (FFT) to classify the fault . Similar approach, to identify and classify the type of fault based on either signal processing techniques namely Fourier Transform, Prony Analysis, S‐Transform, wavelet transform (WT) and Phase Reconstruction or with combination of intelligence method such as support vector machine (SVM), Particle Swarm Optimization (PSO)‐based Artificial Neural Network (ANN), adaptive neuro fuzzy interference system (ANFIS), fuzzy logic system and Fault index method is presented in . The extraction of feature for all these methods to locate the fault is obtained by sampling either the fault current or voltage waveform.…”

Application of ACF-wavelet feature extraction for classification of some artificial PD models of power transformer

“…This issue motivated academics to develop artificial expert systems capable of working with databases and extracting useful information using data mining approaches. Generally, to achieve PD source identification, a knowledge base is derived from raw data by feature extraction methods like PRPD patterns [11,12], FFT [13], statistical analysis [14], cepstral features [15], or wavelet patterns [16,17], and then a decision making system such as neural networks [18,19], SVM [20], PSO [19,21], the Bayes theorem, ANFIS [22], k-means and fuzzy c-means [23], or SOM [24] interprets the knowledge base in a meaningful way to discriminate different PD sources. Methodologies depend on the application and equipment under monitoring; introduction of a new knowledge base still helps experts to make decisions with greater reliability.…”