Digital Twin Development - Condition Monitoring and Simulation Comparison for the ReVolt Autonomous Model Ship

Abstract

Digital Twin technology is becoming an integral part of simulation, testing and operation of semi and fully autonomous vehicles. This technology shows great potential for decreasing testing time, improving cost efficiency and decreasing environmental impact of autonomous vehicles. This thesis includes development towards a digital twin system for DNV GL's concept test ship, ReVolt.

Several practical elements of the digital twin system have been implemented. The digital twin system now includes encoders which measure the speed of the stern motors, a 4G capable boat and a more user friendly visualization in Unity. Field tests were conducted with the objective of collecting crucial data from ReVolt. This data is crucial to test and refine the various digital twin use cases.

The first use case explored is condition monitoring. Several condition monitoring algorithms are researched and two algorithms have been developed and implemented. These produced results that can inform if one of the motors experiences a fault.

The second use case explored was system identification. Data collected from the ReVolt Model during the test day has also been compared with the ReVolt Simulator. Comparisons were made both with respect to the thruster dynamics and the overall boat dynamics. System identification theory is discussed as this can be used to improve the models in the ReVolt simulation.