Just Another Attention Network for Remaining Useful Life Prediction of Rolling Element Bearings

Abstract

Rotating equipment often fails due to faults in the rolling element bearings. The remaining useful life (RUL) prediction of the bearings plays a critical role in prognostics and decision-makers. In this study, attention mechanism is integrated into the internal of just another network (JANET) unit, and a new improved version of JANET unit, namely, just another attention network (JAAN), is firstly presented. Firstly, root mean squares (RMS) of the test-to-failure datasets are calculated to characterize the degradation behavior of the bearings. Then, the prediction model is constructed by stacking multiple JAAN units to estimate the RUL values of rolling element bearings by existing RMS values. Extensive experiments on PRONOSTIA dataset are carried out to validate the superiority of the presented approach. The experiment results show that the proposed JAAN achieves good prediction results than other advanced technologies.

Keywords

• Just another attention network
• remaining useful life prediction
• deep learning
• rolling element bearings