Early fault prediction of DCAE-RNN rolling bearing based on Bayesian optimization
LU Huaicong
LU Tao
LIU Kunhua
QIN Weiwei
GUO Xianglong
HU Jie
Abstract：[Objective]Because the problems of the lack of prior fault knowledge,weak early fault characteristics and strong noise interference in the working environment,it is difficult to accurately predict the early fault of rolling bearings under actual working conditions.A prediction model of denoising and contraction autoencoder(DCAE)combined with Bayesian optimization and recurrent neural network(RNN)was proposed.[Methods]Firstly,the data was rearranged and the data set was divided.The stack de-noising autoencoder was optimized by using Bayesian optimization function,the stack de-noising autoencoder model was trained,the contraction autoencoders was optimized by hyperparameters,and the output of the stack de-noising autoencoder was trained as the input of the contraction autoencoders to get the depth autoencoder.A depth autoencoder was used to extract fault characteristics from vibration signal data of rolling bearing.Secondly,recurrent neural network was used to predict the early fault of rolling bearings.Finally,the XJTU-SY dataset was used for test verification.[Results]The test results show that the proposed method can not only ensure the prediction accuracy,but also reduce the prediction time and computing resource consumption,and improve the deployability of the model.
Keywords：Rolling bearingEarly fault predictionBayesian optimizationDeep autoencoderRecurrent neural network
Publication Date：2026-04-15
Online Publishing Date：2026-09-12(First online date of this platform, not the publication date of the document)
Pages：11( 143-153 )