Hidden fault diagnosis method of digital twin system for crane

The traditional fault diagnosis approach for cranes often results in unclear control status, protracted fault location times, prolonged equipment downtime, and other related challenges. To address these issues, this paper proposes embedding the stacked denoising auto-encoder, multi-class support vector machine, and fuzzy Bayesian network (SDAE-MCSVM-FBN (SMF)) hidden cascade fault diagnosis method into a digital twin model of the crane. The approach utilizes the digital twin to monitor the crane’s state and identify virtual environment faults to make decisions to reduce the direct operation of complex equipment and enhance operational efficiency. This paper first constructs a digital twin model of the crane to describe its composition and interaction behavior within the system. Next, it solves the hidden cascade faults in cranes using the SMF method. Finally, this approach develops a prototype system for a digital twin of cranes with an example in a large steel mill used for verification and analysis purposes. Experimental results demonstrate that compared with traditional manual inspection methods, this paper’s proposed hidden cascade fault method reduces repair time by 24.5%–32.8% while decreasing downtime by 20.5%–32.4%. The outcome highlights the effectiveness of our diagnostic approach and demonstrates digital twin technology can minimize frequent maintenance requirements by personnel while protecting crane service life overall.