Spoke 9 – Models, Methods, Computing Technologies for Digital Twin
A Digital Twin is defined as a virtual replica of a real asset. In order to be coherent with its real counterpart, a lot of effort has to be devoted to the analysis of its operational behavior. For these reasons, data comprehension is fundamental at an early stage, as well as an accurate understanding of the asset functioning, to reach the most accurate artificial intelligence models.
Once the system modeling has been designed, constant communication on both sides of the twin must be maintained. This implies real-time assimilation of data coming from the asset, which allows the improvement and the verification of the models implemented, reducing the error between the DT and its companion. For the achievement of the best replica, we can take advantage of two important tools. The first one is Uncertainty Quantification, providing a quantitative measure of the distance between the twin and the real asset output, while the second one is represented by Reduced Order Models. True real-time feedback can in fact be obtained just from an online response of the model, on time scales compatible with the real asset operational conditions.