Predicting hydrodynamic forces on heave plates using a data-driven modelling architecture

Abstract

Prior work has modelled heave forces on surface piercing bodies in real time for 2D, convex and static geometries using a Hammerstein framework. In this work, the modifications needed to make that framework suitable for a representative heave plate geometry in irregular waves are investigated in order to facilitate force prediction for heaving wave energy converters. Two candidate models are compared. The first introduces a Bouc–Wen model to the conventional internal force calculation in the Hammerstein framework. The second further modifies the dynamic block by replacing the linear ARX dynamic calculation with a nonlinear Kolmogorov–Gabor polynomial (KGP) structure. The addition of the Bouc–Wen model introduces velocity and acceleration dependence and a new hysteresis contribution to the internal force predictions to capture the form drag and inertial loads known to occur around heave plates from the entrained fluid. This addition reduces the average error in the internal force calculation by up to 70% for kinematically controlled bodies in irregular waves. Using the revised internal force in the system dynamic block identification process, it was found that the resulting models provided significant improvements (up to 38%) in prediction of the hydrodynamic heave force when using both the linear and nonlinear ARX formulations.