Advanced diagnosis strategy for incipient stator faults in synchronous reluctance motor

• Published in: 2015 IEEE 10th International Symposium on Diagnostics for Electrical Machines, Power Electronics and Drives (SDEMPED) pp. 110 - 116
• Main Authors: Bouchareb, I., Bentounsi, A., Lebaroud, A.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.09.2015
• Subjects: Ambiguity plane time-frequency representation; artificial neural networks; Circuit faults; classification of faults; detection; Feature extraction; optimal time-frequency; particle swarm optimizer; Reluctance motors; stator winding conditions; Stator windings; synchronous reluctance motor; Time-frequency analysis; Training
• DOI: 10.1109/DEMPED.2015.7303677

This paper presents a new diagnosis method related on the detection and the classification of stator faults in synchronous reluctance motor (SynRM) drives using parametrical and a non-parametrical Time-Frequency Representation (TFR) from a time-frequency ambiguity plane associated with automatic classification method to upgrade the diagnosis field of SynRM. For this purpose, diagnostic features are extracted from current measurements based on non-parametrical Time-Frequency Representation. Then, a feature selection method using Fisher's Discriminant Ratio (FDR) is applied for maximizing the separability between classes with Particle Swarm Optimizer (PSO) technique to select an optimal number of the extracted features which define the feature space. The determination of the decision criterion based on a neural network classifier is carried out for monitoring different operating modes. The detection is realized by parametric TFR based Wigner-Ville Distribution (WVD). Multi-Fault stator winding conditions through the current measurements were used in the classification of short-circuit and open-circuit of axially laminated rotor SynRM. A set of fault scenarios, between healthy, single and combined faults, in terms of current at different load level is taken into account in order to deduce the fault severity. The experimental results prove the efficiency of TFR-ANN method in condition monitoring of electrical machines independent from de type of load and type of machine.