Abstract

Recently, growing concern for the reliability of ship structures has emerged among stakeholders in the maritime sector. In the transition towards predictive maintenance, Structural Health Monitoring (SHM) has emerged as a viable option. SHM systems seek to infer the existence or level of structural degradation using response data from in-situ sensors. In this work, damage detection in marine stiffened panels is treated. Out-of-plane deflections of the panel’s plate elements act as the damage case of interest, due to being potential triggers for buckling collapse. Strain response data are employed as the damage-sensitive features. Two different approaches are employed for damage detection. In the first, the problem is cast as one of multi-class classification, where the different classes correspond to discrete deflection levels, and a detection theory-based classifier is employed. In the second, the problem is treated as one of probabilistic regression, where strain readings are mapped to the probability distribution of the deflection magnitude. Both methods are implemented on a realistic stiffened panel geometry subjected to operational variability, based on an existing vessel, while all the employed data is obtained through Finite Element (FE) simulations. Results indicate that the proposed architecture can effectively provide uncertainty-informed predictions of out-of-plane deflection levels.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call