An Heuristic Approach to a Ship Nonlinear Model Identification

In this paper an identification method based on Genetic Algorithms is proposed as an heuristic approach to the ship model identification. The physical model of the motion of 6 DOF (degrees of freedom) floating object is derived. The ship motion equations are then formulated as first order ordinary differential equation system. Next, the state and output equations of the ship are established. Created model is exploited in the identification process for preliminary verification of proposed identification method. The method explores widely the search space to find the solution and it is resistant to convergence to local minima, which is a main problem in currently used deterministic algorithms. Identified model is a key to conduct advanced research and simulations that allow testing innovative solutions without the additional costs associated with building prototypes and conducting sea trials. Verified ship model can be a useful tool for vessel navigation and control, especially in case of larger vessels, in which response time to changes in control settings is relatively long.