Share Email Print

Proceedings Paper

Fault classification approaches using empirical mode decomposition and histogram of power signals
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

This paper presents two new algorithms for fault classification in power signals. The first algorithm is based on empirical mode decomposition (EMD) of the power signals which decomposes a signal into intrinsic mode functions (IMF). In the proposed technique we obtain the IMFs of the power signals and compute the higher order statistical parameters of each IMF, and a dictionary of feature vectors of different types of faults is prepared. To classify the fault in a given signal, its feature vector is computed and its classification is done using the nearest neighbor rule using its Euclidean distance with the feature vectors stored in the dictionary. The simulation results show that we are able to classify the faults accurately using HOS based approach even at signal-to-noise ratio (SNR) value of 10 dB, which is much lower than the values of SNR reported in the literature. The second method is based on computing the histograms of different types of fault signals and computing their distances with histograms of signals stored in the dictionary. It is observed that above SNR value of 30 dB, we are able to classify all types of faults accurately and this method is computationally less demanding.

Paper Details

Date Published: 28 January 2013
PDF: 7 pages
Proc. SPIE 8760, International Conference on Communication and Electronics System Design, 87601O (28 January 2013); doi: 10.1117/12.2012322
Show Author Affiliations
K. K. Sharma, Malaviya National Institute of Technology Jaipur (India)
Abdul Samad, GLA Univ. (India)

Published in SPIE Proceedings Vol. 8760:
International Conference on Communication and Electronics System Design
Vijay Janyani; M. Salim; K. K. Sharma, Editor(s)

© SPIE. Terms of Use
Back to Top
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?