Ramesh Konatala |
Research Associate, German Aerospace Center (DLR), Institute of System Dynamics and Control, 82234, Weßling, Germany. | Reiko Müller |
Research Associate, German Aerospace Center (DLR), Institute of System Dynamics and Control, 82234, Weßling, Germany. | Marc May |
Research Associate, German Aerospace Center (DLR), Institute of System Dynamics and Control, 82234, Weßling, Germany. | Gertjan Looye |
Head of Department, German Aerospace Center (DLR), Institute of System Dynamics and Control, 82234, Weßling, Germany. | Erik-Jan van Kampen |
Associate Professor, Delft University of Technology, Faculty of Aerospace Engineering, Kluyverweg 1, 2629 HS Delft, Netherlands. |
|
Abstract:
Unforeseen faults during flight can lead to Loss of Control In-Flight (LOC-I), a significant cause of fatal aircraft accidents worldwide. Current offline synthesized, model based flight control methods have limited capability to adapt to unforeseen situations. From a fault tolerance perspective, the Incremental Approximate Dynamic Programming (iADP) controller serves as an ideal model-agnostic, online adaptive control method. This method integrates an online identified locally linearized incremental model with a Reinforcement Learning (RL) based optimization technique, to minimize an infinite horizon quadratic cost-to-go. A key challenge which limits the adoption of these self-learning control methods for flight control is V&V through flight testing. This study addresses the problem by exploring tools, methods and framework for V&V of the online adaptive Flight control law on a CS-25 class Citation-II passenger aircraft. These flight tests mark world's first demonstration of an online RL based automatic Flight Control System (FCS) for this aircraft category, demonstrating real-time learning and adaptation capabilities.
|