Abstract

In this pa,per, a new detection system for early stage short circuit fault in stator winding of induction motor is proposed. The early stage of stator winding short circuit is represented by a low magnitude current and a very short duration that is defined as temporary short circuit. The proposed method is based on transient current recognizing when short circuit fault starting occur and cleared. The transient current during fault is recognized by high frequency signal energy trending of wavelet transform. Three energy of high frequency signal from three consecutive current signal sampling are used as detection variables. Three wavelet types and five levels transformation are evaluated using linear discriminant analysis (LDA) to get the most suitable wavelet transform. The El- man neural network is designed as detection system,. The proposed method is applied, to laboratory experiment. As a result, the proposed method can clearly detect the temporary short circuit fault even though the fault has very fast occurrence and the current magnitude is lower than full load current, with the good accuracy and the ability to provide time information of fault, the proposed method is suitable for monitoring system,.

Original languageEnglish
Pages (from-to)2277-2293
Number of pages17
JournalInternational Journal of Innovative Computing, Information and Control
Volume10
Issue number6
Publication statusPublished - 1 Dec 2014

Keywords

  • Digital signal processing
  • Discrete wavelet transforms
  • Fault detection
  • Induction motors
  • Linear discriminant analysis
  • Recurrent neural networks
  • Short circuit currents
  • Stators
  • Wavelet coefficients

Fingerprint

Dive into the research topics of 'Neural network based real time detection of temporary short circuit fault on induction motor winding through wavelet transformation'. Together they form a unique fingerprint.

Cite this