TWIN generator: AI-process to build a digital twin
Reference number | |
Coordinator | Linköpings universitet - Department of Management of Engineering |
Funding from Vinnova | SEK 2 558 558 |
Project duration | November 2023 - February 2025 |
Status | Ongoing |
Venture | Advanced digitalization - Enabling technologies |
Call | AI for advanced digitalization, 2 |
Important results from the project
The objectives were largely achieved by developing and integrating AI and digitization techniques to create a digital twin. The project provided increased knowledge about data collection, ML methods for component identification and uncertainty management. Key results include insight into data needs, development of synthetic training data, a learning environment at the Visualization Center, and student participation through thesis and project.
Expected long term effects
The project is expected to lead to improved AI-based identification of components in digital twins, which can streamline industrial processes. Furthermore, synthetic training data can reduce the need for manual data collection. The learning environment at the Visualization Center increases knowledge about digital twins, and student participation contributes to skill development. In the long term, this can promote digitalization and automation in several industries.
Approach and implementation
The project combined AI and digitization techniques to create a digital twin from 3D scanning. The work included data collection, evaluation of ML methods for component identification and development of techniques to handle uncertainties. The project was tested on two facilities with varying data quality. In addition, methods for synthetic training data and a learning environment for visualizing the results were developed and exhibited at the Visualization Center in Norrköping.