Motor Bearing Fault Diagnosis Based on Stochastic Resonance Enhanced Stator Current Signals

Due to its simple structure and high reliability, induction motors have become a necessary element in the modern production process. Analyzing the current and vibration signals of induction motor is the primary approach for machine condition monitoring and fault diagnosis. Bearing is an important rotating part of induction motor, and its health status directly affects the running state of the motor. The non-intrusively measured stator current contains a wealth of information about the running state of the motor, and a fault will cause a new frequency component to appear in the stator current. However, due to the effects of noise in the signal and the weak nature of the faulty signal, it is hard to observe these components directly from the original signal. Stochastic resonance is an effective technique to enhance weak signals with noise, and it has been widely used in feature enhancement of rotating machinery faults. Therefore, this paper proposes a second-order underdamped tristable stochastic resonance (UTSR) method based on the characteristics of current signals in order to enhance the fault feature of the induction motor bearing. Firstly, the theoretical signal-to-noise ratio (SNR) of the output signals from the model is examined under the excitation of different sinusoidal components added with different levels of Gaussian noise. It shows that the UTSR model can effectively enhance the fault features. Finally, experimental results show that the UTSR method effectively enhance the bearing early fault characteristics in stator current of the motor.