Real-Time Parameter Estimation of a Nonlinear Vessel Steering Model Using a Support Vector Machine
Journal article, Peer reviewed
Accepted version
Åpne
Permanent lenke
http://hdl.handle.net/11250/2640461Utgivelsesdato
2019Metadata
Vis full innførselSamlinger
- Institutt for marin teknikk [3410]
- Publikasjoner fra CRIStin - NTNU [37304]
Originalversjon
10.1115/1.4043806Sammendrag
The least-square support vector machine (LS-SVM) is used to estimate the dynamic parameters of a nonlinear marine vessel steering model in real-time. First, maneuvering tests are carried out based on a scaled free-running ship model. The parameters are estimated using standard LS-SVM and compared with the theoretical solutions. Then, an online version, a sequential least-square support vector machine, is derived and used to estimate the parameters of vessel steering in real-time. The results are compared with the values estimated by standard LS-SVM with batched training data. By comparison, a sequential least-square support vector machine can dynamically estimate the parameters successfully, and it can be used for designing a dynamic model-based controller of marine vessels.