In this work we present the software architecture used to implement a ship movement prediction system based on a deep learning model. In previous works of the group we recorded the movement of several cargo vessels in the Outer Port of Punta Langosteira (Spain) and created a deep neural network that classifies the vessel movement given the vessel dimensions, the sea state and weather conditions. In this work we present the architectural design of a software system that allows to deploy machine learning models and publish the results it provides in a web application. We later use this architecture to deploy the deep neural network we have mentioned, creating a tool that is able to predict the behavior of a moored vessel 72Â h in advance. Monitoring the movement of a moored vessel is a difficult and expensive task and port operators do not have a tool that predicts whether a moored vessel is going to exceed the recommended movements limits. With this work we provide that tool, believing that it could help to coordinate the vessel operations, minimizing the economic impact that waves, tides and wind have when cargo vessels are unable to operate or suffer damages. Although we use the proposed system architecture for solving a particular problem, it is general enough that it could be used for solving other problems by deploying any machine learning model compatible with the system.
CITATION STYLE
Alvarellos, A., Figuero, A., Sande, J., Peña, E., & Rabuñal, J. (2019). Deep Learning Based Ship Movement Prediction System Architecture. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 11506 LNCS, pp. 844–855). Springer Verlag. https://doi.org/10.1007/978-3-030-20521-8_69
Mendeley helps you to discover research relevant for your work.