At present, research dedicated to the development of the Maritime Autonomous Surface Ship (MASS) faces a new round of challenges in terms of safety and intelligence. In real-world application scenarios, it is crucial for a vessel to navigate safely and have a status awareness to prevent any collision during vessel encounter situations. The study proposes a comprehensive framework along with the models for decision support of the vessel operation using data science analytics. The models are built with descriptive historical dataset which will be processed and analyzed through the machine learning approaches. Several marine operation aspects including vessel movement, external forces, and vessel machinery were taken into account to increase the reliability and accuracy of models.
This thesis includes four ancillary chapters (Chapter 1-3, 6) and also two study manuscripts (Chapter 4-5) to introduce and demonstrate the new approaches to the decision support model for vessel operation, employing a modern data science technique. The first study manuscript exhibits the decision support model for vessel machinery maintenance while the second study exhibits the decision support model for vessel navigational operation, specifically based on the water area. Both study manuscripts were focused along the South Korean sea, utilizing the dataset around the period of 2019~2020.
Using data analysis techniques, such as stepwise regression and multivariate polynomial regression, the first study manuscript obtained the variable association model. The model proposes a new approach to detect the frequent failures of a ship's equipment and/or machinery through its corresponding variable, once the model found low to moderate-strength relations between the target response and the corresponding variable.
The second study manuscript aimed to introduce navigation traffic element features for decision support models of vessel anomaly behavior. Utilizing both AIS and GIS data to look for each boating characteristic, the vessel behavior could be defined clearly, whether it is sailing in a normal state, close to an anomalous behavior state, and/or in an anomalous state.