Multi-Channel Vibro-Acoustic Fault Analysis of Ball Bearing using Wavelet based Multi-Scale Principal Component Analysis

Ball bearing fault segmentation at different time steps are important to avert failure. This paper studies the Vibro-acoustic characteristic of the ball bearing using Wavelet Based Multi Scale Principal Component Analysis (WMSPCA) and FFT. Firstly, the characteristic frequencies of the ball bearing for healthy and unhealthy states are verified using an impulse exciter hammer; and the generated frequencies are acquired using a Zigbee wireless accelerometer sensor. Secondly, the acoustic and vibration characteristics are acquired using three channel accelerometer sensor and a array microphone. Lastly, the actual characteristics of the ball bearing are extracted using WMSPCA. The main advantage of WMSPCA lies in the actual feature segmentation from different channels independent relative to the direction of propagation of faults. WMSPCA uses wavelet and PCA to auto-correlate and cross-correlate the signal simultaneously. The algorithm extracts the frequency range of operation of the ball bearing and assists in determining the precise frequency of vibration excluding its perplexed frequency components associated along tangential, axial and radial direction of the ball bearing. The paper also correlates the significance of acoustic-vibration in the fault finding of bearing.