Μελέτη του εκτιμώμενου χρόνου άφιξης πλοίων στο λιμάνι του Πειραιά με την χρήση νευρωνικών δικτύων
A study on ships’ estimated time of arrival at the port of Piraeus using neural networks
KeywordsΝευρωνικά δίκτυα ; Automatic Identification Systems (AIS) ; Estimated Time of Arrival (ETA) ; Εκτιμώμενος χρόνος άφιξης ; Μοντέλο πρόβλεψης ; Πειραιάς
The 1978 SOLAS Convention stipulates in Regulation 19 / Chapter V the navigation equipment to be carried on board on ships, according to their type. In 2000, the International Maritime Organization (IMO) adopted a new requirement (as part of a revised new chapter V) for all ships, to have Automatic Identification Systems (AIS). This system aims at better and more effective control of maritime traffic by the competent authorities since the ship's signals contain both dynamic information such as vessel's position and static data such as vessel's identity. The Datastories Lab of the University of Piraeus has installed an antenna for recording AIS signals and creating a corresponding database. In addition, the knowledge of Estimated Time of Arrival (ETA) is important for better management and organization of port operations. In this thesis, this time variable will be studied. The study and estimation of this time is done by using a neural network, which will take as input a series of data and as output will give the estimated time of arrival of the ship. Input data consists of AIS signals stored at the database of the university. The evaluation of the result shall be based on the actual time of arrival of the ship and the deviation from it. Finally, we will study which factors can affect the outcome, such as the weather conditions or the time horizon of the data we have.