Reaming Useful Life Prediction of Bearings Based on CNN and Encoder Layers

In the era of Industry 4.0 of smart manufacturing, the study of the remaining useful life (RUL) of machinery and equipment has rightly become the focus of researchers. As a key component of Prognostic Health Management (PHM), immediate RUL prediction can minimize economic losses and improve safety. This paper proposes a model combining convolutional neural networks (CNN) with encoder layers for RUL prediction of bearings. In this model, CNN with encoder layers is used for extracting degenerate feature and then a linear layer is introduced to solve the RUL regression prediction of the bearings. The experimental validation is based on the IEEE PHM Challenge 2012 dataset. Comparing with the experimental results of other existing methods, the RMSE of this paper is superior, which confirms the correctness of the model of this paper.