Digital twin concepts for linking live sensor data with real-time models

Abstract

For taking most advantage of live or real-time sensor measurements, data have to be processed by a single or even by a chain of models on the fly, in contrast to earlier offline simulation solutions. This requirement can be best met by concepts developed under the general term “digital twin” (DT). The step from the Internet of Things (IoT) to a full exploitation of DT solutions entails new challenges but also provides new features, which we discuss based on our example DT solution for remote monitoring of fruit during ocean transportation. A crucial challenge is the transformation of models into an updateable format, necessary to keep the physical object and its modelled representation in sync. A basic new feature of DTs is new software solutions for easy and flexible linking of different models through a streaming platform by implementing an event-driven architecture. We demonstrate a solution for controlling model execution during multiple life cycle phases of the fruit as physical object. An evaluation of response times showed that server performance is sufficient to handle more than 100 DT instances per second.