CIS Digital Twin Times 2021 | 15 Nov. 2021 | Lausanne Switzerland
Prof. Karen E. Willcox, Director, Oden Institute for Computational Engineering and Sciences, University of Texas, Austin
Predictive Digital Twins: From physics-primarily based modeling to scientific machine learning
Abstract
A digital twin is an evolving digital design that mirrors an individual actual physical asset throughout its lifecycle. Critical to the digital twin thought is the potential to perception, collect, review, and master from the asset’s facts. To make digital twins a reality, several aspects of the interdisciplinary area of computational science, like physics-based mostly modeling and simulation, inverse troubles, uncertainty quantification, and scientific machine learning, have an essential position to enjoy.
In this get the job done, we acquire a probabilistic graphical product as a official mathematical illustration of a digital twin and its associated physical asset. We produce an abstraction of the asset-twin system as a established of coupled dynamical programs, evolving above time by means of their respective state-areas and interacting by means of observed information and management inputs. The abstraction is recognized computationally as a dynamic final decision community. Predictive abilities are enabled by physics-centered lowered-buy designs. We reveal how the approach is instantiated to generate, update and deploy a structural digital twin of an unmanned aerial automobile.
