A Digital Twin of a Web Processing Machine Using CAD, AI and Vision.

Presented by Juan Chowdhury, Ghent University

Co-Authors: Juan Chowdhury, Yentl Thielemans, Arul Mathivanan, Jeroen D.M. De Kooning, Kurt Stockman, Guillaume Crevecoeur

Computer-Aided Design (CAD) models are information-rich and can be quickly upgraded to motion simulations to evaluate the operation and performance of machines. However, in realistic industrially relevant mechatronic systems, these motion simulations fail to achieve the desired realism and do not incorporate the process executed by the machine. The modeling of such complex behavior is strenuous and difficult to validate. CAD-enriched vision systems improve state observability which is modelled by hybrid artificial intelligence algorithms. These interventions rely on a robust IT network to exchange data in real time. The result is a Digital Twin of the physical system.