Development of a ship autopilot using neural network

Gouletas, Stefanos; Γουλέτας, Στέφανος

dc.contributor.author	Gouletas, Stefanos	en
dc.contributor.author	Γουλέτας, Στέφανος	el
dc.date.accessioned	2020-09-28T11:14:00Z
dc.date.available	2020-09-28T11:14:00Z
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/51177
dc.identifier.uri	http://dx.doi.org/10.26240/heal.ntua.18875
dc.rights	Default License
dc.subject	Autopilot	en
dc.subject	Naval	en
dc.subject	Control system	en
dc.subject	Neural network	en
dc.subject	Artificial intelligence	en
dc.subject	Σύστημα ελέγχου	el
dc.subject	Τεχνητή νοημοσύνη	el
dc.subject	Αυτόματος πιλότος	el
dc.subject	Ναυπηγική	el
dc.subject	Νευρωνικά δίκτυα	el
dc.title	Development of a ship autopilot using neural network	en
heal.type	bachelorThesis
heal.classification	Control system engineering	en
heal.language	en
heal.access	free
heal.recordProvider	ntua	el
heal.publicationDate	2020-02-12
heal.abstract	This Thesis concerns the development of a neural network autopilot which will be used for vessel controlling. The main purpose of this project is to use alternative and more sophisticated technologies like artificial intelligence in contrast with the commonly used PID controllers. The Neural Network controller was constructed using Python libraries (Tensorflow and Keras), which have been used for the easier and more efficient training and building. For the representation of the ship system, we used two different mathematical models, one linear (Nomoto Model) and one more complicated based on the hydrodynamic coefficients, in order to make the system more realistic (4DOF Model). For testing, we executed a simple course keeping simulation to check the suitability of the controller. Even then in order to observe the efficiency of the neural network controller, we use it in a path following scenario in which the vessel should follow a predefined course. The performance of the neural network controller is compared to classical PID control results which are implemented for the same simulation scenarios. We conclude that artificial intelligence and more specifically Neural Networks is a very important tool which could be used for more efficient and effective controllers that could process more informations as sea state, weather and fuel optimization.	en
heal.advisorName	Papalambrou, George	en
heal.committeeMemberName	Spyrou, Kostas	en
heal.committeeMemberName	Papadopoulos, Christos	en
heal.academicPublisher	Εθνικό Μετσόβιο Πολυτεχνείο. Σχολή Ναυπηγών Μηχανολόγων Μηχανικών. Τομέας Ναυτικής Μηχανολογίας	el
heal.academicPublisherID	ntua
heal.fullTextAvailability	false