Abstract

Induction motors are widely used in various industrials applications due to their performance and strength however there occurs some faults such as shaft improper alignment, bearing damages, rotor bar malfunction. This paper presents the solution for handling such defects and detecting them with the help of vibration signals from the motor parts. This method incorporates signal processing, machine learning based divisions. Here a Butterworth filtering and wavelet denoising module to remove the noise and preventing from other faults occurring in the motor. Statistical and spectral features are used with the help of Principal Component Analysis. The obtained features are then classified with Support Vector Machine and Random Forest Algorithms. Simulated and real time vibration involved makes the proposed system achieve high fault detection accuracy under various circumstances. This study provides a low cost, efficient solution for maintenance in industry environments.