Vmd-Cnn-Gru Based Fault Diagnosis Method for Rolling Bearings Under Variable Operating Conditions

Le résumé fourni par la source

To address the challenge of diagnosing rolling bearing faults under variable operating conditions and noise, this paper presents a neural network-based combined model for effective fault diagnosis. The model employs the Sparrow Search Algorithm (SSA) to optimize key parameters in Variational Modal Decomposition (VMD) and integrates a Convolutional Neural Network-Gated Recurrent Unit (CNN-GRU) for fault classification. Firstly, for the vibration signals of rolling bearings in different states, the penalty factors and modal components in the VMD are optimized using SSA, and the optimal modal components are extracted by decomposition through the VMD so as to maximally preserve the essential features of the signals and reduce the noise interference. Subsequently, a deep learning network comprising a convolutional layer, a batch normalization layer, a ReLU activation layer, and a max pooling layer is designed to extract spatial features at various scales; finally, the time series data after the extracted features are processed by the gated recursive unit (GRU) to capture the timing information and complete the fault category diagnosis. In this paper, the Case Western Reserve University (CWRU) dataset is used to validate the proposed model. By comparing the experiments with other models, the results show that the model has significant advantages in both classification accuracy and operational efficiency, and can be effectively applied to the rolling bearing fault diagnosis task, which is especially suitable for industrial scenarios with variable operating conditions and complex noise.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.