Support Vector Machine Learning in Marine Hydrodynamic

Abstract

The development is proposed of Support Vector Machine learning algorithms in marinehydrodynamics. They include the forecasting of s"hip responses in a seastate, the identification ofcomplex viscous and nonlinear free-surface flow physics affecting the responses o"f vessels incalm water and in steep waves and the use of SVM algorithms in ship design. The forecasting ofvessel responses will be based on nonlinear SVM regression methods using seastate elevationsand vessel response records as ~features~. Viscous forces affecting ship responses arise from flowseparation around bilge keels and the hull. The identification of these forces will be carried outby SVM nonlinear regressions functions of flow ~features~ identified by the supervised learningof the SVM algorithm against experimental measurements.

Document Details

Document Type
DoD Grant Award
Publication Date
Sep 29, 2017
Source ID
N000141712985

Entities

People

  • Paul Sclavounos

Organizations

  • Massachusetts Institute of Technology
  • Office of Naval Research
  • United States Navy

Tags

Readers

  • Marine Hydrodynamics
  • Neural Network Machine Learning.

Technology Areas

  • AI & ML