Data-driven digital twin of hybrid power systems

Data-driven digital twin of hybrid power systems

Background: Artificial Intelligence/neural network have recently been widely implemented in different fields for modelling and predicting the behavior of the system. Implementation of AI based models in marine system will open different possibilities. Moreover, abundant data harvesting is implemented in new systems not only to monitor the system, but also to analyze the data and their trends. Use of these data also in modelling and model training adds a step towards the development of both autonomous and digital twin of the system. In addition, efficient model development based on data may also help in real time simulation of the complex systems.

Objective: The objective of the project is to study different methods applicable to data-driven modeling and use of data set for model tuning of the hybrid power components. Then, the method will be applied to model a case study hybrid power system such as a typical diesel engine-generator or fuel cell. The AI algorithm will be established to generate a data-based model of the engine. To model the engine, the necessary data for different variables and parameters of the engine can be collected by running Kongsberg simulator (K-Sim Engine simulator) in different scenarios. For the case of fuel cell, the data can be collected from the Hybrid Power Lab. Finally, explore the pros and cons on the use of AI and data-driven models in commercial simulators.

This task will be performed in connection with Open Simulation Platform (OSP) and in close collaboration with Kongsberg.

Collaborator: Kongsberg Digital