Ship maneuvering model optimization for improved identification with less excitation

Abstract

The precise model of a ship is the foundation for its operation and control. The traditional methods of building such models are time-consuming and require large amounts of simulation data, sea trial experiments, or calculating with professional computational fluid dynamics software. Usually, specific maneuvers are conducted to obtain the experimental data with full excitation. In this paper, we propose a ship maneuvering model optimization method that can lower the requirement on data excitation during model identification by simplification. First, the least squares method is used to identify the preliminary parameters of the ship mathematical model. Then, correlation analysis can determine the correlation among the parameters and divide the parameters with higher correlation into one group. Sensitivity analysis is used to detect the influence level of parameters and as a basis for selecting the more critical parameters. Based on the results of these two analyses, we set up a standard to simplify the ship maneuvering mathematical model. Finally, the simplified model and the complete model are tested under different levels of data excitation, and the experiment results verify that the simplified model can perform better than the complete model when identifying with less excitation data.