Health Indicator Construction of Rolling Bearings Based on Deep Convolutional Neural Network Considering Phase Degradation

摘要

Health indicators (HI) attempt to identify and quantify history and ongoing degradation processes by extracting feature information from collected data. The main problems in the existing HI construction methods are as follows: (1) existing methods use fixed learning rate to train deep learning network, which is inefficient; (2) existing HIs construction methods do not take into account phase degradation; (3) failure threshold is difficult to determine in remaining useful life (RUL) prediction. In this paper, a HI construction method of rolling bearings based on a deep convolutional neural network with polynomial decaying learning rate (PDCNN) considering phase degradation is proposed. In this paper, the original vibration signal is input into the PDCNN, and the features extracted from the deep convolutional neural network are input into the deep neural network (DNN) to construct HI. The training label setting takes into account phase degradation. The polynomial decaying learning rate is used to train the neural network efficiently. Examples show that the HI constructed by the proposed method is superior to compared three methods in monotonicity and trendability. The polynomial decaying learning rate can improve the training efficiency of deep neural network and save a lot of time. The value of the constructed HI is in the interval [0, 1], which solves the problem that the failure threshold is difficult to determine in RUL prediction.