Digital twins as research tools and research objects in aviation
Aircraft and their components are extremely complex systems. During their development, innumerable data sets and models are created concerning their behaviour and interaction. In the coming years, digital twins will increasingly be employed in this context as images of physical components (in this case: turbine, wings, etc.) and will gain significantly in importance. As a result, development processes will become leaner and new digital business models will emerge.
In the DigECAT project (Digital Twin for Engine, Components and Aircraft Technologies), digital twins of various aircraft components of an aircraft system are being realised under operating conditions using the DLR research aircraft ISTAR as an example. The two central research questions of the digital twin as a research tool and the digital twin as a research object can also benefit other projects: These are projects that are characterised by mutual methodological developments, as well as research that uses the digital twin as a data source. Consequently, connections to more than 10 further planned or ongoing DLR projects are possible. DigECAT will run from 01.12.2022 to 31.12.2025. A total of 12 DLR institutes from the fields of aeronautics, materials science and software and IT are involved. The activities are essentially divided into four main work packages:
(1) Software Methods: Here, methods, procedures, standards and tools for the utilisation and development of digital twins are being developed. Particular attention is being paid to the further development of the data management and access architecture STASH, which was developed in the preceding project DigTwin and which logs all DLR research flights and renders the measurement data accessible. Furthermore, the integration of in-depth provenance and metadata - such as time, pilot information, and details of the aircraft - is also being undertaken.
(2) Wing Moveables: The entire life cycle of an aircraft component is mapped in a digital twin. Within the project, a focus on the feedback from the operation is being incorporated into the design (DevOps cycle). At the end of the project duration, coupling with the physical aircraft ISTAR is planned.
(3) Engine: A digital comparison process between design and real geometries is being conceived. The end-to-end process for the evaluation of the geometries is being developed, and various technologies for the digitalisation of real geometries are being investigated and compared with one another. The result is a digital twin at component level. The overall process is an essential building block for the engine concept of the ISTAR digital twin.
(4) ISTAR: The aircraft will be connected to the STASH architecture developed in DigTwin and will perform research flights.
Contribution Institute for AI Safety and Security
The researchers analyse state-of-the-art architectures, standards and components from the GAIA-X context - such as data security, data sovereignty, and data availability - and customise them for the project. The focus here lies on IDS and GAIA-X Federation Services.
Within the work package Data Provenance and Metadata, the AI Institute focuses on the conception of a demonstrator for the data-processing operations for metadata and ontology management. The constant flow of real-time data results in the challenge that digital twins also have to undergo permanent change. The goal is to ensure the continuous integration of data into automated processes in order to enable data-quality management or analysis and model derivation using methods of machine learning or artificial intelligence.
The researchers are adapting the DLR internally developed data-management system Shepard (Storage for heterogeneous product and research data) in order to meet the project requirements, thereby enabling provenance data to be functionally integrated into semantic technologies (ontology approaches). In this context, investigations are being carried out into how privacy-preserving technologies or concepts from the field of zero-knowledge proofs can be integrated into a distributed data ecosystem in order to handle data with high protection requirements. The Institute for AI Safety and Security is hereby supporting the participating institutes in the implementation.