In this paper, condition monitoring of induction machines using air-gap magnetic flux density spectrum via artificial neural networks is presented. The proposed scheme is chosen due to its effectiveness, simplicity, and low cost that used for the detection of broken rotor bar faults. The spectrum of the air-gap magnetic flux density is estimated using the Fast Fourier Transform, which can capture the fault related to harmonic components. The extracted information is then utilized by a machine-learning paradigm in a Multi-class classification approach for the detection of broken rotor bars, for both, adjacent and non-adjacent using artificial neural networks as a classification method. The obtained simulation results of the healthy and faulty conditions using finite elements prove the applicability of the proposed method.
展开▼
