Roller Bearing Defect Detection With Multiple Sensors

Three levels of bearing outer race damage were tested over varying rotational speed and radial load. An acoustic emission sensor (AE) and accelerometer were used to record acoustic emission and vibration levels, respectively. The high frequency resonance technique (HFRT) and Adaptive Line Enhancer (ALE) were employed to process vibration signals. The primary metric of vibration level in this research is the Root-mean-square (RMS). The HFRT and ALE signal processing combination is effective in isolating deterministic defect frequencies. Finally a multi-layer perceptron neural network using accelerometer and AE RMS values was demonstrated as being successful in bearing defect level classification given operating conditions.