Logo image
Back
Bayesian Hankel Extended Dynamic Mode Decomposition for System Identification of High-Speed Planing Hulls
Conference proceeding

Bayesian Hankel Extended Dynamic Mode Decomposition for System Identification of High-Speed Planing Hulls

Giorgio Palma, Andrea Serani, Matteo Diez, Christian Milano, Zhaoyuan Wang, Sungtek Park, Deniz Ozturk Sarigul and Frederick Stern
SNAME International Conference on Fast Sea Technology, FAST 2025
2025
DOI: 10.5957/FAST-2025-071

View Online

Abstract

This study explores Bayesian Hankel extended Dynamic Mode Decomposition with control (BHeDMDc) as data-driven, model-free methods for predicting the response of the Generic Prismatic Planing Hull (GPPH) in wave conditions. This approach decomposes complex vessel dynamics into spatial-temporal coherent modes, incorporates time-delay embeddings and extended observables for improved robustness to transient and nonlinear effects, integrates control inputs to enhance predictive accuracy, and includes probabilistic uncertainty quantification. The method is applied to towed motion in irregular head waves. Performance metrics include motion variables and forces reconstruction. Analysis shows that BHeDMDc effectively captures dominant dynamic features with real-time predictive capability essential for digital twin applications. The method is capable of addressing the direct problem, predicting motions from wave signals, and the inverse problem, predicting forces acting on the hull from motion variables easily measurable on board. This research highlights the strengths of the method and supports the development of uncertainty-aware, data-driven models for high-speed naval vessels, with potential extensions toward real-time adaptation and validation using experimental data.
 Data-Driven Dynamic Mode Decomposition Generic Prismatic Planing Hull High Speed Planing Craft Irregular waves Loads Motions Nonlinear Prediction System Identification Uncertainty

Details

Metrics

1 Record Views
Logo image