S. Chakravorti

Estimation of time-to-flashover characteristics of contaminated electrolytic surfaces using a neural network

A major field of neural networks (NN) application is function estimation, because the useful properties of NN such as adaptivity and nonlinearity are well suited to function estimation tasks where the equation describing the function is unknown. In this paper the prerequisite training data are obtained from experimental studies performed on a flat plate model for a polluted insulator under power frequency voltage. Detailed studies have been carried to determine the NN parameters which give the best results. Studies have also been carried out to assess the effect of the presence of inadequate data in the training set on modeling accuracy. It is found that, when training is completed, NN is capable of estimating the function t = f(V,L 1 R p ) very efficiently and effectively even when the inadequate data are incorporated in the training set.

Cross-wavelet transform as a new paradigm for feature extraction from noisy partial discharge pulses

In this work a new approach based on cross-wavelet transform towards identification of noisy Partial Discharge (PD) patterns has been proposed. Different partial discharge patterns are recorded from the various samples prepared with known defects. A novel cross-wavelet transform based technique is used for feature extraction from raw noisy partial discharge signals. Noise is a significant problem in PD detection. The proposed method eliminates the requirement of denoising prior to processing and therefore it can be used to develop an automated and intelligent PD detector that requires minimal human expertise during its operation and analysis. A rough-set theory (RST) based classifier is used to classify the extracted features. Results show that the partial discharge patterns can be classified properly from the noisy waveforms. The effectiveness of the feature extraction methodology has also been verified with two other commonly used classification techniques: Artificial Neural Network (ANN) based classifier and Fuzzy classifier. It is found that the type of defect within insulation can be classified efficiently with the features extracted from cross-wavelet spectra of PD waveforms by all of these methods with a reasonable degree of accuracy.

Boundary element studies on insulator shape and electric field around HV insulators with or without pollution

Capacitive-resistive field computations are carried out around post-type HV insulators of varying shapes. The boundary element method (BEM) has been employed for electric field computations. Different insulator shapes have been obtained by varying several parameters, which define the shape of the HV insulator contour. For each insulator shape, the maximum stress occurring on the insulator surface has been determined with no surface pollution, uniform surface pollution and also partial surface pollution. For partial pollution, several cases have been studied, in which different sections of the insulator surface are polluted. Furthermore, the effect of electrode radius on the maximum stress on insulator surface has been investigated. The results obtained are presented in this paper in detail.

Wavelet-aided SVM tool for impulse fault identification in transformers

Accurate diagnosis of faults in transformers can significantly enhance the safety, reliability, and economics of power systems. In the case of a fault, it has been established that the pattern of the fault currents contain a typical signature of the nature and location of the fault for a given winding. This paper describes a new approach using wavelet transform (WT) for extraction of features from the impulse test response of a transformer in time-frequency domain and support vector machine in regression mode to classify the patterns inherent in the features extracted through the WT of different fault currents. This paper also describes an approach to identify the type and location of the transformer faults accurately by analyzing experimental impulse responses that contain noise. Here, experimental impulse responses have been preprocessed with the help of wavelet-packet filters to remove the unwanted noise from the signal and thereby enhance the analyzing capability of continuous wavelet transform.

A methodology for identification and localization of partial discharge sources using optical sensors

The present work represents a methodology to detect the location of single as well as multiple Partial Discharge (PD) sources by optical method and to investigate the performance of optical sensors for this purpose. An experimental setup has been arranged in the laboratory for generation of PDs, optical sensing and analysis of the recorded signals obtained from multiple optical sensors. The analysis results prove the effectiveness of the methodology using optical sensors to find whether PD is occurring at single location or multiple locations. For identification of PD locations pattern recognition technique has been utilized by considering the received optical energy as a feature. For feature selection and classification two techniques have been evaluated, viz. Gaussian Mixture Model (GMM) and Support Vector Machine (SVM), and both have shown promising performance. SVM in regression mode was used for identification of unknown PD location/locations. In this case average accuracy obtained was 92.6% when PD is occurring at one location and 80.1% when PD is occurring at two locations. The obtained results indicate that, the proposed methodology can be used to locate partial discharges in high voltage equipment where the optical signals due to discharges find a path to get radiated towards the outer surface.

Application of artificial neural networks for optimization of electrode contour

In this paper artificial neural networks (NN) with supervised learning are proposed for HV electrode optimization. To demonstrate the effectiveness of artificial NN in electric field problems, a simple cylindrical electrode system is designed first where the stresses can be computed analytically. It is found that once trained, the NN can give results with mean absolute error of /spl sim/1% when compared with analytically obtained results. In the next section of the paper, a multilayer feedforward NN with a back-propagation algorithm is designed for electrode contour optimization. The NN is first trained with the results of electric field computations for some predetermined contours of an axisymmetric electrode arrangement. Then the trained NN is used to give an optimized electrode contour in such a way that a desired stress distribution is obtained on the electrode surface. The results from the present study show that the trained NN can give optimized electrode contours to get a desired stress distribution on the electrode surface very efficiently and accurately. >

Rough-granular approach for impulse fault classification of transformers using cross-wavelet transform

A novel approach based on information granulation using Rough sets for impulse fault identification of transformers has been proposed. It is found that the location and type of fault within a transformer winding can be classified efficiently by the features extracted from cross-wavelet spectra of current waveforms, obtained from impulse test. Results show that the proposed methodology can localize the fault within 5% of the winding length with a high degree of accuracy. The basic concepts of feature extraction using cross-wavelet transform and the method of classification of those features by rough-granular method are also explained.

Wavelet analysis of optical signal extracted from a non-contact fibre-optic vibration sensor using an extrinsic Fabry–Perot interferometer

Interferometric optical fibre sensors have proved to be many orders of magnitude more sensitive than their electrical counterparts, but they suffer from limitations in signal demodulation caused by phase ambiguity and complex fringe counting when the output phase difference exceeds one fringe period and for multiple fringes. This paper presents a novel signal decoding technique based on the wavelet transform of optical data extracted from a non-contact vibration sensor using an extrinsic Fabry-Perot interferometer (EFPI) implemented using single-mode fibre. The EFPI cavity has been used to generate an optical interference signal between two parallel, highly reflective surfaces separated by a variable distance. Firstly, a few recorded experimental results of the interference fringe formation due to vibration are presented in this paper. Then the wavelet transform is used for decoding the vibration signature for three major purposes of the data analyses, namely elimination of noise from the optical signals collected in real time, identification of the frequency breakdown points of the signal efficiently and automatic counting of the interference fringes. In turn, the wavelet transform is successfully employed to decode the vibration signature from the non-stationary output signal of an EFPI sensor.

Monitoring of inter-turn insulation failure in induction motor using advanced signal and data processing tools

Detection of stator winding inter-turn insulation failure at early stages is crucial for promoting safe and economical use of induction motors in industrial applications. Whereas major insulation failures involving larger percentages of winding are easily discernible from magnitude of supply current, minor inter-turn insulation failures involving less than 5% of turns often go undetected. The present contribution reports experimental results of minor faults due to inter-turn insulation failures in stator windings of induction motor under different loading conditions being analyzed using data and signal processing tools combining Parks Transform and Cross Wavelet Transform. Rough Set Theory (RST) based classifier has been used for fault severity monitoring.

Wavelet aided SVM classifier for stator inter-turn fault monitoring in induction motors

Early detection of faults in stator winding is crucial for reliable and economical operation of induction motors in industries. Whereas major winding faults can be easily identified from supply current magnitude, minor faults involving less than 5% of turns are not readily discernible. The present work documents experimental results for monitoring of minor short circuit faults in stator windings of induction motor. Motor line current has been analyzed using modern signal processing and data reduction tools combining Parks transformation and Continuous Wavelet Transform (CWT). Support Vector Machine (SVM) based data classification tool has been used for fault characterization based on fault features extracted using CWT.