Skip to Main Content
The aim of this paper is to present a wavelet-based method for broken-bar detection in squirrel-cage induction machines. The frequency-domain methods, which are commonly used, need speed information or accurate slip estimation for frequency-component localization in any spectrum. Nevertheless, the fault frequency bandwidth can be well defined for any squirrel-cage induction machine due to numerous previous investigations. The proposed approach consists in the energy evaluation of a known bandwidth with time-scale analysis using the discrete wavelet transform. This new technique has been applied to the stator-current space-vector magnitude and the instantaneous magnitude of the stator-current signal for different broken-bar fault severities and load levels.