Prediction Method of Bearing Remaining Useful Life Based on PEWA-Transformer

The prediction of the remaining useful life(RUL) of bearing is a key technology to ensure the safe operation of mechanical equipment. However, the long sequence and periodic characteristics of bearing vibration signals make it a great challenge to effectively predict the remaining service life. The existing residual service life prediction methods often only consider this problem from the perspective of long-term series relationship capture, and ignore the attention of sudden fault characteristics, which may lead to the neglect of key degradation information, thus affecting the accuracy of prediction. At the same time, the adaptability of existing methods is also relatively general when the time span of bearing data increases, which will significantly increase the amount of calculation while reducing the accuracy of prediction. In order to solve the above problems, this paper proposes an improved bearing remaining useful life prediction method PEWA-Transformer. By adding Contextual Position Encoding and Window attention mechanism, the ability to capture dependent features and extract local features in long time series is enhanced, and the accuracy of residual life prediction is improved. The experimental results show that the model can accurately predict the remaining service life of bearings.