The TwinRotor project conducted between 2017 and 2019 aimed to create a digital twin of a rotor system. A proof of concept digital twin system was conceived during the project and a data driven machine learning model for the dynamic behaviour of the rotor was created. Further applications of similar data driven methods are virtual sensors utilizing collected data from a fleet of installed products, which can improve condition monitoring and predictive maintenance services.
Embedded sensors and actuators combined with modern networking, cloud, and machine learning technologies have made it possible to collect and analyze massive amounts of data reflecting the use of industrial products. This data explosion provides obvious opportunities to optimize the operation of products and systems in terms of energy consumption, material usage, or quality control. Collecting data from a fleet of installed products can improve condition monitoring and predictive maintenance services as well as further value adding services.
In the research project the aim was to improve the behavior of rotating machinery using a digital twin coupled with Industrial Internet methods to support enhanced data flow between the machinery, simulation based virtual sensors, and applied big data analytics. This will lead to insights into how the rotating machinery design can be improved, in addition to better operational efficiency of the machinery and enhanced quality of the products manufactured with them. The wider scientific objective was to study how Industrial Internet methodologies coupled with machine learning can be applied especially to complex engineering design.
The project Digital Twin of Rotor System was funded by the Academy of Finland and lasted until the end of 2019. The project was conducted together with Lappeenranta University of Technology.
Professor Petri Kuosmanen