Transient signal analysis and classification for condition monitoring of power switching equipment using wavelet transform and artificial neural networks

Abstract

In this paper a transient signal processing technique is developed for condition monitoring. This technique is especially applicable to analysing vibration signals which are produced by switching mechanisms. Multiresolution and wavelet transforms are combined to extract salient features with limited dimension from the primary vibration signals. These features are further classified by artificial neural networks for the purpose of condition assessment. The results provide the foundation for the effective application wavelet analysis to condition monitoring of mechanical switching devices utilised in electricity supply systems.