Student:	Murat Eray KORKMAZ
Supervisor:	Prof. Dr. Ercan KÖSE
Department:	Makine Mühendisliği
Institution:	Graduate School of Natural and Applied Sciences
University:	Karadeniz Technical University Turkey

Title of the Thesis:	SHIP BERTHING BY USING ARTIFICIAL NEURAL NETWORKS
Level:	Ph.D.
Acceptance Date:	14/1/2019
Number of Pages:	173
Registration Number:	Di1294

Summary:

Ship berthing has an important place in ship handling. It is the most complex and

difficult problems of ship handling. Maneuvers performed by the captain s knowledge and

experience, need to control a lot of parameters that are effective. As of there is lots of

uncertainty, these are the most complicated and difficult problems of ship handling.

Artificial neural networks (ANN), which can simulate the biological brain, are

systems that can learn as humane, interpret the data they learn and produce results by using

this information. At this point, ANN has become the most suitable systems for the solution

of these kinds of problems.

Within the scope of this thesis, a neural network model has been designed for a fixed

step propeller container vessel which approached and berthing to Singapore Port by using

artificial neural networks. The berthing maneuvers for different wind conditions have been

carried out by the ship s captain (also a specialist on a full mission bridge simulator). With

the data obtained from that berthing maneuvers, a neural network that using the Multi-Layer

Perceptron model is trained by using the Levenberg-Marquardt learning algorithm. By using

the trained neural network, the ship is provided to perform the berthing maneuver for a

different wind condition that the network never seen. When the results are evaluated, it is

seen that ANN can be used in ship berthing maneuvers.