Performance Evaluation of Nonparametric, Parametric, and the MUSIC Methods to Detection of Rotor Cage Faults of Induction Motors

This paper aims to analyze three different spectral decomposition methods applied to the stator current of induction machines to detect rotor broken bars, namely Welch, Burg, and MUSIC (Multiple Signal Classification). Each of these methods is based on different concepts of power spectral estimation: non-parametric, parametric and eigenvalue decomposition, respectively. The frequency resolution, variance and detection capability are different for each method according to the set of parameters used. The paper also aims to determine which method is best suited for the implementation in automated fault detection systems. The evaluation is based on the sampled current taken on a prototype machine running under different load and faulty conditions. The effect of the main parameters of each method on the capacity to detect faults is also evaluated and compared. The comparison is performed considering the ability to discriminate fault related frequencies in the corresponding power spectrum. Different window types, window length, overlap and sampling frequency were analyzed and compared.