A Model Based Approach for Digital Testbed Development supporting Virtual Experimentation of an Unmanned Surface Vehicle

The idea of creating a digital enterprise that focuses on a virtual testing environment has gained interest as digital systems have become more prominent and integrated in the engineering world. As digital twins become a central focus of the digital enterprise, a growing interest lies in what techniques can be applied to enhance the design, validation, and verification of a system. To take the novel step towards enabling full digital twins and predicting the performance of an Unmanned Surface Vehicle (USV) under untested conditions, an architecting approach is needed. A hybrid model using both physics and data-driven information about a system to capture its behavior over three layers of interest: dynamic, electrical, and thermal and a simulation-based digital testbed was explored to understand how improved testing could reduce uncertainty in model performance. This study is enabled by physical testing of an USV, data regression, modeling and simulation, and requirements management. Earlier implementation results have indicated that using a combination of physics and data driven models as the virtual version of a physical system has great potential in capturing the true performance of the physical twin and enabling accurate system-level virtual testing. It was observed that, despite error and uncertainty accumulation throughout the testing and physics model derivation, a methodology for developing a digital testbed aimed at reducing uncertainty in a future digital twin was successfully identified.