This paper reviews digital twin (DT) research on bearings under marine operating constraints, focusing on modeling strategies, application tasks, and evidence boundaries. Existing studies are grouped into direct marine evidence, marine-adjacent evidence, and generally transferable evidence. Three modeling routes—physics-driven, data-driven, and hybrid-driven—are then summarized for condition monitoring, fault diagnosis, and remaining useful life (RUL) prediction. The review indicates that a preliminary framework for marine bearing DTs has emerged. Physics-driven methods support state reconstruction and mechanism analysis, data-driven methods enable rapid identification and local prediction, and hybrid methods are promising under complex conditions and limited data. However, high-level validation of actual marine bearing systems remains insufficient, especially in terms of long-term online monitoring, shipboard field tests, and full-chain verification for water-lubricated stern bearings. Current conclusions should therefore be regarded as evidence-bounded assessments. Future work should emphasize online updates, consistency between the virtual and physical domains, and hierarchical validation.
Shi et al. (Tue,) studied this question.