Multiple signal classification based on automatic order selection method for broken rotor bar detection in induction motors

Multiple signal classification (MUSIC) algorithm has been widely used to obtain high-resolution frequency estimation for an accurate identification of frequency components in low signal-to-noise ratios. One of the main drawbacks associated with the use of the MUSIC algorithm is that its performance is fully deteriorated when a wrong frequency signal dimension order is chosen, producing that some spurious frequencies could appear or some signal frequencies could be missing. In this paper, it is proposed a multi-objective optimization method to address the frequency signal dimension order problem. The proposed approach is based on a novel feature extraction of frequency components, which allows determining an adequate frequency signal dimension order. The methodology has been integrated as part of the MUSIC algorithm, and it can find the optimal order within a predefined frequency bandwidth, where the user is interested to find a frequency component. To evaluate the effectiveness of the proposed methodology, experimental results from several current signals obtained in the detection of broken rotor bar fault in induction motors have been tested.