Identification of ship manoeuvring motion based on nu-support vector machine

A practical and robust system identification modelling method for ship manoeuvring motion is presented, to alleviate the impact of noise-induced problems, such as parameter drift or over-fitting, on the model reliability. The method is based on v(nu)-Support Vector Machine (v-SVM) algorithm, which can automatically control the number of support vectors to ensure the sparsity of the solution. Multiple standard manoeuvring datasets are simultaneously used as training data to cope with the parameter drift issues. The proposed method is validated by using polluted simulated data of three different levels. The results demonstrate that the identified model has good generalisation, verifying the robustness and efficiency of the algorithm.