Digital twins are virtual representations of physical assets and systems that replicates and extends information about processes and objects in the physical system, by models built on historical and real-time data. Two-way communication between the physical twin and the digital twin enables automated actions and decision making on the physical system.
Data-driven approaches for modelling of the assets and systems enables real-time monitoring of the status, events and anomalies in the physical system, optimization of logistics and process automation, classification and predictive analysis, risk and uncertainty analysis. All these features are further enhanced by the ability to simulate scenarios in the digital twin to understand implications for its physical counterpart.
The implementation and development of digital twins for industrial applications are most successful when data of high quality is available. As the building of digital twins require detailed domain knowledge in addition to data management, architectural design, knowledge representation, physics, machine learning and AI, several stages in developing the digital twin is needed, and off-the-shelf solutions are limited.
