Real-Time Estimation of Bearing Fault in Inverter-Fed Three-Phase Induction Motor

Inverter-fed induction motors are so popular for industrial applications because of their simple and rugged construction, minimum maintenance requirement as well as a possible wide range of speed and torque of the motor drive. Faults in a motor can affect the continued operation of the system which causes loss in terms of quality of product and raw material. The extreme working conditions, dynamic loading of the motor, non– sinusoidal inverter voltages, and vibrations are the main reasons for the development of faults in motors. The early detection of incipient faults is important because of this, scheduled maintenance is feasible, and the production line can be optimally utilized during light load conditions, avoiding the complete shutdown of the plant. This work proposes efficient diagnostic routines based on the stator current signal to identify the rotor bearing faults at an early stage. The stator current signals are acquired from the inverter-fed induction motor drive in real-time and discrete wavelet transform has been applied to the acquired signals to study the spectral components using the approximate and detailed coefficients. The statistical properties of DWT of current signals are used to form feature vector L2-norm. rigorous analysis is used for the feature extraction. The extracted features are given to support the vector machine classifier for bearing fault identification.