Non-parametric dynamic system identification of ships using multi-output Gaussian Processes

• Published in: Ocean engineering Vol. 166; pp. 26 - 36
• Main Authors: Ariza Ramirez, Wilmer, Leong, Zhi Quan, Nguyen, Hung, Jayasinghe, Shantha Gamini
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 15-10-2018
• Subjects: Dependent Gaussian processes; Dynamic system identification; Multi-output Gaussian processes; Non-parametric identification; Oceanic vehicles; Multi-output Gaussian processes; Dependent Gaussian processes; Dynamic system identification; Oceanic vehicles; Non-parametric identification
• ISSN: 0029-8018; 1873-5258
• DOI: 10.1016/j.oceaneng.2018.07.056

A novel application of non-parametric system identification algorithm for a surface ship has been employ on this study with the aim of modelling ships dynamics with low quantity of data. The algorithm is based on multi-output Gaussian processes and its ability to model the dynamic system of a ship without losing the relationships between coupled outputs is explored. Data obtained from the simulation of a parametric model of a container ship is used for the training and validation of the multi-output Gaussian processes. The required methodology and metric to implement Gaussian processes for a 4 degrees of freedom (DoF) ship is also presented in this paper. Results show that multi-output Gaussian processes can be accurately applied for non-parametric dynamic system identification in ships with highly coupled DoF.
•A methodology for the application of multi-output GPs for dynamic system identification of ships has been developed.•Qualities and defects of multi-output Gaussian processes based dynamic system identification are presented.•A Study of Ship dynamic system identification with Multi-output GPs was executed to show the methodology and applicability.