Digital Twin-Driven Intelligent Monitoring of a Marine Gearbox Based on CNN-LSTM Network

High-power marine gearboxes are subjected to complex internal and external excitations brought by the harsh working environment. Therefore, it is indispensable to realize the intelligent operation and maintenance of marine gearboxes. In this paper, a virtual and real state mapping technology is introduced to construct a marine gearbox digital twin system (MGDTS). The overall architecture of the MGDTS is designed to accurately reflect the operating status of the marine gearbox. Based on the CNN-LSTM hybrid neural network, a fault early warning model is established by integrating the physical model and twin data, so as to realize the intelligent monitoring of the marine gearbox. In addition, the twin data and operation status are visually monitored for closed-loop management of the gearbox. Finally, the digital twin results agree well with the ground truth, R2 of transmission efficiency, velocity and displacement respectively are 0.9034, 0.9936 and 0.9984, indicating the excellent performance of the MGDTS.