A Kind of Fault Diagnosis Method of Marine Engine Bearing Based on GWO-Informer-SENet

Marine engine bearings are critical components of ship propulsion systems, and their operational status directly impacts vessel safety and efficiency. During navigation, these bearings perform essential functions, including power transmission and shaft support. Any failure of bearings could result in a complete propulsion system breakdown, potentially leading to serious safety incidents and substantial economic losses. Therefore, effective fault diagnosis for marine engine bearings is of paramount importance. The proposed GWO-Informer-SENet method offers an innovative solution for marine bearing fault diagnosis. This approach integrates the Grey Wolf Optimizer (GWO) with an enhanced Informer model, where GWO automatically optimizes hyperparameters to significantly improve the model’s generalization capability and operational adaptability. The Informer model, a Transformerbased deep learning architecture, demonstrates exceptional time-series modeling capabilities, enabling the effective extraction of complex features and long-term dependencies in bearing operational data. Additionally, the incorporation of the Squeeze-and-Excitation Network (SENet) module further enhances channel-wise adaptive feature learning, substantially improving fault-type identification accuracy and diagnostic efficiency.