Gated Recurrent Unit-Based Neural Network for Enhanced Bearing Fault Diagnosis

Motors are commonly used in industrial applications, where bearing problems frequently occur, posing a significant safety risk to industrial production. Traditional fault diagnostic methods, which often rely solely on signal processing techniques, have proven to be ineffective. To address this issue, deep learning (DL) has rapidly developed and achieved remarkable results in fault diagnosis. This paper proposes an intelligent fault diagnosis and classification method for rolling bearing faults based on Ensemble Empirical Mode Decomposition (EEMD) and a stacked Gated Recurrent Unit (GRU) neural network. The vibration signal is decomposed into several Intrinsic Mode Functions (IMFs) using EEMD to eliminate random noise interference from the original vibration signal. Selecting sensitive features from both the time and frequency domains of IMF components is crucial, and this is achieved by using the correlation coefficient value. Finally, a GRU model is developed to classify faults based on the extracted features. The proposed model accurately classifies different types of faults under real operating conditions and is compared with existing techniques. The method demonstrates superior diagnostic performance, achieving an overall model accuracy of 100%.