Methods which combine engineering methods with machine learning models were developed, in order to make it possible to get more accurate and self-learning models for real-life performance of ships. In the framework of the project, the research was focused on developing suitable models which are applied on measured operational data from the vessel. Using such models requires a lot of computation power and thus a framework has been developed to allow the automatic analysis of the data in cloud environments. The results of the analysis are presented to the end user through a web-based business intelligence tool.
Part of the work was to establish automatic and reliable data communication infrastructure to gather data on board and to transfer it to the shore based system, providing sufficient access for globally distributed user base.
Validation of the results was carried out in cooperation with major organizations in the international maritime field (ship owners, operators, classification societies and shipyards).
With the rise of environmental awareness and economical pressure from the ever increasing cost of fuel, the ship designers and ship operators are in pursuit of optimizing the ship design and ship maintenance program to minimize energy consumption of the ship. Traditionally there has been no means to accurately know the ships performance in real operation. In ship design a fixed, pre-defined “sea margin” has been used to prepare for the increased power need in real operation, without good knowledge of what is the real power need in actual operations. In ship operations, the maintenance program of the vessel has a significant effect in the energy consumption. However, without accurate information about ship’s performance, it is impossible to know if, e.g. hull cleaning should be performed once a month or once a year. With accurate information the optimum maintenance program can be determined.
The tool / system has been taken into full operational use with selected customers of NAPA. The customers have an online access to see the performance data of the ships in almost real time. Any deviations in the performance can be seen and possible reasons discovered with flexible data analyzing tools as part of the system.
Esa Henttinen, NAPA