Bearing Fault Diagnosis Research Based on Empirical Mode Decomposition and Deep Learning

The bearing condition of the mechanical equipment is significant for the equipment. The existing bearing fault diagnosis have the disadvantages of slow running speed and poor accuracy. In response to this deficiency, a fault diagnosis method is proposed, that by using empirical mode decomposition (EMD) and deep learning (DL). DL uses long short-term memory (LSTM) network. EMD breaks down bearing data into intrinsic mode function (IMF) by the bearing vibration signal, and correlation analysis is performed on IMF to collect feature data of bearing data. Then, LSTM network is used to build model and adjust parameter value. This model can identify and classify the extracted feature information to diagnose faults and types of faults. Simulation achievements show the feasibility of the proposed method in bearing diagnosis.