Dynamic Runtime Integration of New Models in Digital Twins

The development of cyber-physical systems is heavily relying on model-driven approaches. After deployment, these models can be utilised in a Digital Twin setting, acting as virtual replicas of the physical components and reflecting the behaviour of the running system in real-time. Complex systems often consist of numerous models interacting with each other and individual models may need to be updated after deployment. This means that new models need to be integrated and swapped during runtime without interrupting the running system. In this paper, we propose an approach for model-based Digital Twins to replace individual models without stopping or halting the operation of a cyber-physical system. Furthermore, our approach allows to replace not only individual models, but also update the overall structure of the interaction of models in the Digital Twin setting. The use of the proposed mechanism is illustrated through two case-studies with an agricultural robot prototype.