Data-Driven Real-Time Decision Support and its Application to Hybrid Propulsion Systems
This paper describes a method for providing real time decision support based on measurements rather than optimizing a mathematical model. The proposed method is thus beneficial for systems for which the modelling would be inaccurate, the dynamics and complexity of the system would make it difficult to optimize in real time, or the risk of returning local minima is not acceptable. The proposed method is implemented on four fishing vessels. These vessels are complex and give the skipper many choices related to how the vessel is operated. The developed tool advises the crew on in real time on operational decisions, particularly on the use of various diesel electric and diesel mechanic propulsion modes, including decisions such as the use use of shaft generator, direct coupling between main engine and propeller or not, propeller pitch, etc. This will presumably reduce both fuel consumption and emissions of CO2 and NOX. Some examples of obtainable results from both onshore analyses and the onboard application are presented to demonstrate the methods applicability.