Fault Diagnosis of Rolling Bearings of Seawater Pumps Based on Deep Residual Shrinkage Network

The seawater circulating pump is a vital equipment in the gasification system of LNG terminals, and its rolling bearings serve as a key component. Therefore, accurately diagnosing bearing faults holds great significance. Given the harsh operating conditions of the seawater circulating pump, where the characteristic frequencies of bearing vibration signals are often masked by environmental vibrations, leading to insufficient accuracy in conventional diagnostic methods, this paper proposes a novel approach for fault diagnosis of rolling bearings in seawater circulating pumps based on a deep residual shrinkage network m Fault Diagnosis of Rolling Bearings of Seawater Pumps Based on Deep Residual Shrinkage Network odel. This model employs deep residual shrinkage modules to mitigate the interference of redundant information such as environmental vibrations during fault feature extraction. By incorporating soft thresholding and an attention mechanism, it reduces feature loss during deep network training and enhances the accuracy of fault classification. Finally, experimental validation demonstrates that under conditions of significant environmental vibrations, this model outperforms other commonly used network models in terms of accuracy and robustness, providing a reliable solution for fault diagnosis of rolling bearings in seawater circulating pumps.