A combined modular parametric and non-parametric method for planar ship motion's on-line prediction

Abstract

A combined modular model is presented for planar ship motion prediction by assembling parametric and non-parametric forecasting method. This parametric module and non-parametric module represent the holistic and local time-varying dynamics of system, respectively. The nonlinear auto-regressive with exogenous input model is implemented for on-line prediction, and variable-structure radial basis function network is employed to represent the time-varying dynamics of system. The experiment of planar ship motion prediction is conducted and the results demonstrate the accuracy and robustness of the presented combined model. © 2011 Springer-Verlag.