What question did this study set out to answer?

The research aims to evaluate the efficacy and compliance of ML-based surrogate models in legacy avionic systems.

February 19, 20260 citationsOpen Access

Implementation and certification of ML-based surrogate models in avionic systems

CGChristophe GabreauAirbus (India)NVNicolas ValotAirbus (France)FPFilipo PerottoOffice National d'Études et de Recherches Aérospatiales

Key Points

The research aims to evaluate the efficacy and compliance of ML-based surrogate models in legacy avionic systems.
Assessment of surrogate models in the context of legacy systems
Evaluation of implementability and certifiability standards
Review of regulatory guidelines from EASA regarding ML applications
Surrogate models are found to be provable and implementable in legacy systems
Results indicate compliance with future EASA guidelines
The potential for improved safety and efficiency in aviation applications is highlighted.

Abstract

This paper argues that surrogate models can be used in legacy systems because they are provable, implementable and certifiable with respect to the future development assurance guidance for ML applications anticipated by the European union Aviation Safety Agency (EASA).

Demander à l'IA

Bookmark

View Full Paper