Remaining Life Prediction Method for Rolling Bearing Based on the Long Short-Term Memory Network

• DOI: 10.1007/s11063-019-10016-w
• Publication Date: 2019-03-18T19:04:53Z
• Authors: Fengtao Wang; Xiaofei Liu; Gang Deng; Xiaoguang Yu; Hongkun Li; Qingkai Han
• Keywords: Feature (linguistics); Similarity (geometry)

A residual life prediction method based on the long short-term memory (LSTM) was proposed for remaining useful life (RUL) prediction in this paper. Firstly, feature parameters were extracted from time domain, frequency domain, time–frequency domain and related-similarity features; then three feature evaluation indicators were defined to select feature parameters that could better represent the degradation process of bearings and constructed the feature set with the time factor. The data of the feature set was used to train the LSTM network prediction model, and then the RUL was predicted by the trained neural network. The full life test of rolling bearing was provided to demonstrate that this method could accurately predict the remaining life of the rolling bearing, and the result was compared with the prediction results of BP neural network and support vector regression machine to verify the effectiveness.