An algorithm for offline identification of ship manoeuvring mathematical models from free-running tests

•An identification algorithm for ship manoeuvring mathematical models is developed.•The procedure is based on minimizing a distance between the reference and recovered time histories•The distance was measured using 5 different metrics including the Hausdorff metric.•Validation of the algorithm is carried out using simulated responses artificially polluted•Using the Hausdorff metric it was possible to reach the robustness of the identification algorithm.

An identification algorithm for ship manoeuvring mathematical models has been developed. The identification procedure is based on the classic genetic algorithm used for minimizing a distance between the reference and recovered time histories. The distance was measured using 5 different metrics including the Hausdorff metric. Validation of the algorithm was carried out using simulated responses artificially polluted with the white noise of various levels. It was demonstrated that by only using the Hausdorff metric it was possible to reach necessary robustness of the identification algorithm.