Indirekte Berechnung von Robustheits- und Performance-Garantien für adaptive Regler
Zusammenfassung der Projektergebnisse
Flight control software is a highly safety-critical component in aviation. And as such, it must undergo a rigorous verification and validation process, to prove operational reliability. To meet this requirement, flight control systems must have high robustness characteristics. But for traditional control laws, there is a fundamental trade-off between performance and robustness. And there are metrics such as gain and phase margins that can be used to quantify robustness and guarantee safe operations. An approach that allows to keep high performance and necessary robustness is to use adaptive control. However, given the nonlinear and non-deterministic nature of adaptive controllers, the known robustness and performance metrics cannot be applied. The main goal of the project "Indirect computation of robustness and performance guarantees for adaptive controllers" was to develop a method to guarantee robustness and performance in an adaptive controller. For this research, a model of a F-16 aircraft was adapted from open source data. This model was then incorporated in the trimming and linearization, and Monte Carlo simulation frameworks developed at the institute of flight system dynamics. Then a baseline controller was developed using the Incremental Nonlinear Dynamics Inversion (INDI) concept. The baseline controller was augmented with an adaptive controller based on the predictor MRAC. The adaptive controller was implemented in a plant augmentation architecture, which allows to separate the baseline controller from the adaptive controller. Initially, the method of guarantee of robustness was to be developed such that guarantees are provided off-line. A robustness theorem was therefore developed using the gap metric. Given the highly mathematical aspect of the gap metric, a collaboration was initiated with a mathematician, who is a Professor at the Texas A&M University in the United States. The resulting robustness theorem was used to apply boundaries on the adaptive controller. And it was mathematically proven that if these boundaries are satisfied, then the adaptive controller would not harm the robustness and performance of the proven baseline controller. However, this approach was very conservative in such a way that the system could be stable even when the boundaries were violated. There was a need to develop a method for the computation of the gap metric for nonlinear systems. But the arrival of the COVID-19 pandemic made the collaboration with the Professor from the Texas A&M University impossible. Therefore, the work on this approach was stopped, and a new approach was developed. The new approach was aimed at providing runtime safety assurance for the adaptive controller. This means there would be a monitoring system that would deactivate the adaptive controller if it starts diverging from the expected behavior. In this regard, four monitoring concepts were developed. The last concept presented in the report is still being investigated, and the results are to be published in a near future. The goals of the project could thus be achieved, although the methodologies used were different from what was originally planned. The monitoring concepts allow to provide guarantees of the operational safety of the adaptive controller, and they are currently being investigated further in the framework of two ongoing student theses. And these methods might be a stepping stone towards certification of adaptive controller.
Projektbezogene Publikationen (Auswahl)
-
Adaptive Augmentation of Incremental Nonlinear Dynamic Inversion Controller for an Extended F-16 Model. AIAA Scitech 2019 Forum. American Institute of Aeronautics and Astronautics.
Bhardwaj, Pranav; Akkinapalli, Venkata Sravan; Zhang, Jiannan; Saboo, Saurabh & Holzapfel, Florian
-
Modeling and Incremental Nonlinear Dynamic Inversion Control for a Highly Redundant Flight System. AIAA Scitech 2019 Forum. American Institute of Aeronautics and Astronautics.
Zhang, Jiannan; Wang, Jian; Zhang, Fubiao & Holzapfel, Florian
-
Robustness of Adaptive Control Augmentation of Linear Infinite Dimensional Systems Using the Kato Gap Metric. AIAA Scitech 2020 Forum. American Institute of Aeronautics and Astronautics.
Balas, Mark J.; Jaisle, Jerg & Holzapfel, Florian
