A new dynamic model of electromagnetic coupling in induction motors with dynamic eccentricity

Induction motors are extensively used in renewable energy systems, manufacturing, and electric vehicles, making early-stage condition monitoring of dynamic eccentricity crucial to ensure its reliable and safe high-speed operation while preventing severe failures. Experimental studies have shown that control voltage significantly affects the characteristics of dynamic eccentricity faults and the amplitude of their associated harmonics. This influence limits the accuracy of current-based fault diagnosis. However, most existing electromagnetic coupling models do not consider the effects of control voltage on the dynamic behaviour of stator current. Consequently, the underlying mechanism governing the motor’s frequency response under dynamic eccentricity—especially when subjected to realistic control voltage excitation—remains inadequately understood. To address this problem, this paper proposes a novel multiple coupled circuit (MCC) model that simultaneously considers the time-varying mutual inductance induced by rotor eccentricity and realistic pulse width modulation (PWM) excitation from an open-loop inverter. By leveraging modified winding function theory, the model enables accurate computation of inductances and stator current response under varying eccentricity and load conditions. The model has been verified by the designed custom-built experimental test rig. Comparative analysis with experimental data confirms that the proposed model accurately reproduces both time-domain waveforms and spectral features of stator current, including eccentricity-induced sidebands masked by PWM harmonics. Results indicate that PWM excitation elevates the spectral noise floor, which obscures low- frequency fault components; however, modulation features around the principal slot harmonics remain detectable. The study further reveals that the amplification of sideband increases approximately linearly with the severity of eccentricity. They are also affected by load torque (due to slip variations) and by the carrier frequency of the PWM control, which modulates the spectral distribution of fault-related components. These findings clarify the interplay between control voltage, mechanical load, and fault-related spectral features, providing a solid foundation for effective condition monitoring under realistic operating conditions.