Noise Assessment in the Diagnosis of Rolling Element Bearings

In this paper we perform a noise analysis to assess the degree of robustness to noise of a neural classifier aimed at performing multi-class diagnosis of rolling element bearings. We work on vibration signals collected by means of an accelerometer and we consider six levels of noise, each of which characterized by a different signal-to-noise ratio ranging from 40.55 db to 9.59 db. We classify the noisy signals by means of a neural classifier initially trained on signals without noise, then we repeat the training process with signals affected by increasing levels of noise. We show that adding noisy signals to the training set we manage to significantly increase the classification accuracy.