Information interval spectrum: A novel methodology for rolling-element bearing diagnosis

Abstract

As a popular methodology in bearing fault diagnosis, envelope analysis also has limitations because it looks no more skilled at diagnosis of weak faults in heavy noise environments. For trying to overcome these disadvantages and improve the effectiveness of diagnosis, a novel bearing fault diagnosis methodology called as information interval spectrum (IIS) was put forward here. Vibration signals are first divided into frames and the fluctuation feature is extracted from each frame by a feature operator. The fault interval is then defined by the threshold processing of each feature frame and transformed into fault impulse to intensify its recognition effects. Faults are identified in spectrum. Our methodology is verified by means of simulations and experiments. The corresponding results indicate that our IIS can diagnose weak faults in low signal-to-noise ratio, prevent heavy attenuation of the fault harmonics in spectrum, and eliminate the interference of the rotation frequency.