A Linear Algebraic Based Diagnosis Method for Broken Rotor Bars of Line Start Permanent Magnet Synchronous Motors

Abstract

Line start permanent magnet synchronous motors (LSPMSMs) combine the high efficiency of the permanent magnet synchronous motors (PMSM) with the ease of use, simplicity in design and high starting capability of induction motors (IM). Due to the rapidly growing usage of this relatively new motor, studying its performance under fault conditions is necessary. In this thesis, the coupled magnetic circuit and winding function approach have been used to develop a mathematical model for LSPMSM under broken bar fault condition. This model takes into account the rotor asymmetry due to the broken bar fault in the qd reference frame. The effects of the broken bars fault on the rotor resistances and inductances have been evaluated. Motor’s torque, rotor speed and stator current signatures of the LSPMSM have been captured using MATLAB/SIMULINK® for both healthy and faulty cases with different numbers of broken bars at different loading levels. Results have been verified by comparison with the JMAG® FEM model results, where high level of agreement has been obtained. In addition, this thesis presents an artificial neural network (ANN) method for detecting the broken rotor bar faults based on using singular value decomposition (SVD) extracted from the stator phase current. The accuracy of the proposed diagnostics algorithm reaches 96% when applied to stator current signals of the motor under unseen load and broken bar conditions.