Real-time eccentricity detection of PM motors by monitoring speed and back EMF induced in an additional winding

Abstract

The technique to detect a motor fault before motor failure is an important topic for maintaining the reliability of motors and motor-driven systems. Bearing faults are major motor faults, because the bearings are the most vulnerable and flexible components of the motor. Some signals are generated by the motors before motor failure, including signals of a rotor’s dynamic and static eccentricities. In this dissertation, a method to detect the dynamic and static eccentricities of a permanent magnet (PM) motor is developed by monitoring a fault detection signal in real time without performing any further post-processing, even under a non-stationary rotational speed. The mathematical equation of the back electromotive force (EMF) induced in a tooth-coil winding is derived and various cases are simulated to detect the dynamic and static eccentricity of a PM motor. Finally, the fault detection signal is proposed to detect the dynamic and static eccentricities in real time without performing any further post-processing, even under non-stationary rotational speed. The two-dimensional finite element (FE) model is developed to verify the proposed method of a three-phase PM motor with 8 poles and 12 slots. The experiment setup is also developed to verify the proposed method. The fault detection signal can successfully detect the dynamic eccentricity in a PM motor in real time. First, the background of the dissertation is presented, which covers the motor faults of induction and PM motors. The previous work on condition monitoring of induction and PM motors is presented. Second, the method to detect the dynamic and static eccentricities of a PM motor by monitoring the back EMF induced in a tooth-coil winding is presented. The amplitude sum of the sideband frequency components of the back EMF induced in a tooth-coil winding is utilized to detect the dynamic eccentricity of a PM motor. The amplitude of the main frequency component of the back EMF induced in a tooth-coil winding is utilized to detect the static eccentricity of a PM motor. Third, this dissertation investigates the method to detect the dynamic eccentricity of a PM motor by monitoring a fault detection signal in real time without performing further post-processing even under non-stationary rotational speed. The fault detection signal is defined as the back EMF in an additional winding divided by rotational speed, and the additional winding is wound around the teeth corresponding to the even number of pole pitches. This research confirms that the dynamic eccentricity generates the fault detection signal induced in the additional winding. The peak-to-peak amplitude of the fault detection signal induced in the additional winding increases linearly with dynamic eccentricity. Therefore, the dynamic eccentricity can be identified by monitoring the amplitude of fault detection signal induced in the additional winding, regardless of rotational speed, without performing any further post-processing. Finally, this dissertation proposes the method to detect the static eccentricity of PM motor in real-time without further performing any post-processing. The proposed method to detect the static eccentricity uses the back EMF induced in an additional winding and rotational speed of rotor. The fault detection signal is defined as the back EMF induced in an additional winding divided into rotational speed. The additional winding is wound around the teeth located on the opposite side of the motor with the opposite winding direction to estimate the difference of the back EMF between two teeth. This research confirm that the static eccentricity generates the fault detection signal, and the peak-to-peak amplitude of the fault detection signal increases linearly with increase of static eccentricity. Therefore, the static eccentricity can be identified by monitoring the amplitude of fault detection signal induced in the additional winding, regardless of rotational speed, without performing any further post-processing.