The maintenance of aircraft and their components is subject to complex and safety-critical requirements. Airplanes are designed for decades under extreme conditions, and the loads of their sometimes high-priced components are correspondingly high. In order to minimize failure risks and maximize the lifespan, regular component tests are required, in which extensive data sets arise.
So far, this test data for error detection and localization has mostly been evaluated and manually. As a result, valuable information from past diagnoses, test courses and component knowledge remain largely unused. At the same time, limited data availability in rare errors is a challenge for established diagnostic methods.
In the ProMoDi research project, it is therefore examined how data-driven diagnostic models can be developed, which can be physically interpreted and systematically link test data with domain-specific previous knowledge. The goal is to automatically recognize error conditions, to narrow down and to interpret them in their cause, even under uncertainty and with limited database.
For this purpose, both real and synthetically generated test data are used to train AI-based models that can be adapted to different system architectures and components. This includes the simulation of technical disorders as well as the data-based learning of physical dependencies within the assembly. In this way, diagnostic models are to be created that not only classify errors, but also capture their causality and systemic effects, such as by taking error dependencies between individual components. The modeling and learning procedures used are further developed in the course of the project and extended to new technical and data-related requirements. This creates a flexible, scalable foundation for intelligent diagnostic systems in aircraft maintenance.
For practical implementation and validation, a model-based research environment is built up with the aim of integrating them into real test benches. It serves as a test platform for new diagnostic procedures and as an interface between theory, simulation and maintenance practice.
Project duration: 01.01.2021 – 31.12.2026
The project is funded in the DTEC funding program by the Federal Ministry of Defense (BMVg).
Letzte Änderung: 20. June 2025