A framework and application for dynamic assessment of ship equipment based on multi-source data fusion and digital twins

Addressing diagnostic challenges in complex marine environments (high humidity, high salinity, multi-source interference, diverse fault types), this study uses a marine centrifugal fan to construct an intelligent condition assessment framework integrating digital twin and multi-source heterogeneous sensing. Vibration, visual, and speed data are synchronised and mapped to a lightweight digital twin for dynamic state perception. A fusion strategy based on robust Mahalanobis distance and convolutional neural networks identifies six fault modes (weighted F1 = 0.9965). A visual detection model tailored for ship compartments achieves mAP@0.5 of 0.861 (cracks) and 0.827 (corrosion). Multi-modal results enhance adaptability and robustness to offshore conditions. Interactive programs enable real-time monitoring and fault visualisation. Experiments confirm high diagnostic accuracy, fast response, and efficient deployment, supporting intelligent ship equipment condition management.