Ensuring the dependability of Cyber-Physical Systems (CPSs) poses challenges for model-based engineering, stemming from the semantic heterogeneity of the models of computational, physical and human processes. We argue that delivering such dependability requires a marriage of multi-disciplinary multi-models developed and co-simulated during design with models derived from real operational data. The assets developed during design thus become a learning digital twin, able to support machine-assisted decision-making both in redesign and in responsive operation. Starting from an integrated open toolchain for CPS design, we illustrate the potential for integrating diverse discrete-event and continuous time models with inductive models derived from such operational data in a digital twin setting.

Peter Gorm Larsen is professor at the Department of Engineering at Aarhus University, where he both leads the software engineering research group as well as the Centre for Digitalisation, Big Data, and Data Analytics (DIGIT). After receiving his MSc in EE at the Technical University of Denmark in 1988, he worked in industry before returning to complete an industrial PhD in 1995. In his industry career, as a development engineer and manager, he gave industrial courses all over the world. He returned to academia in 2005. His research is on the tools that help engineers to leverage models in early design stages in particular for the development of CPSs. He is the author of more than 150 papers published in journals, books and conference proceedings.