TCN-CBAM-BiGRU-based Fault Diagnosis Method for Rolling Bearing of Marine Engine

The rolling bearing is crucial for load transmission and motion support in marine engines, and their fault diagnosis is of great significance for ensuring the navigation safety of ships and extending the service life of equipment. In response to the issues of insufficient sensitivity to features, unidirectionality, and inadequate capture of shortterm temporal relationships in bearing fault detection using Time Series Convolutional Network (TCN) networks. This article proposes a fault diagnosis model based on TCN-CBAM BiGRU. This model utilizes a TCN to extract the temporal features of bearing vibration signals, and weights the temporal features through a Convolutional Block Attention Module (CBAM) to enhance the model’s attention to key features. On this basis, the bidirectional gated recurrent unit (BiGRU) is used to further extract the dependency relationships and contextual information of the time series. The experiment shows that the model has good diagnostic performance and anti-interference ability, and can be used for fault diagnosis tasks of marine engine bearings.