Comparison of approaches for parametric sea state estimation using the wave-buoy analogy and neural networks on full-scale measurement sata

Monitoring the surrounding sea state enhances operational safety and energy efficiency of vessels during voyage. The directional sea state can be estimated from measured vessel responses using physics-based models following the Wave-Buoy Analogy (WBA) or data-driven methods like Convolutional Neural Networks (CNNs). This study explores both approaches for parametric sea state estimation using simulation and full-scale measurement data of a barge. The issue of local minimum convergence in the least-squares algorithm is addressed by a modified implementation of the parametric WBA. The modification involves constraining the relative wave direction during optimization, leading to more robustness against outliers in the estimation of the peak period and relative wave direction. The study emphasizes that the methods provide feasible approaches for parametric sea state estimation on the given data, with the WBA requiring knowledge of the vessel-specific conditions and the CNN depending on the representativeness of the training data.

Subjects

Sea state estimation

wave-buoy analogy

full-scale measurements

response amplitude operators

DDC Class

623.8: Naval Architecture; Shipbuilding

006.32: Neural Networks

More Funding Information

This study has been conducted using E.U. Copernicus Marine Service Information; DOI: 10.48670/moi-00022

Lizenz

https://creativecommons.org/licenses/by/4.0/

Publication version

publishedVersion

Name

full_iwsh2025-p0058.pdf

Size

3.86 MB

Format

Adobe PDF

Options

Comparison of approaches for parametric sea state estimation using the wave-buoy analogy and neural networks on full-scale measurement sata