Publication Details

Reliability-Based Control System Optimization in Uncertain Conditions

NOVÁK Jiří, HANÁK Jiří and CHUDÝ Peter. Reliability-Based Control System Optimization in Uncertain Conditions. In: Las Vegas, 2024.
Czech title
Optimalizace řídicího systému založená na spolehlivosti v nejistotním prostředí
Type
conference paper
Language
english
Authors
Novák Jiří, Ing. (DCGM FIT BUT)
Hanák Jiří, Ing. (DCGM FIT BUT)
Chudý Peter, doc. Ing., Ph.D. MBA (DCGM FIT BUT)
Keywords

Polynomial Chaos Expansion, Cross-Entropy Method, Model Predictive Control

Abstract

This paper presents an automated control system tuning approach with emphasis to reliability with respect to vehicle's defined Operational Design Domain (ODD). A joined approach based on Cross-Entropy Method (CEM) and Polynomial Chaos Expansion (PCE) based Kriging meta-model is used to iteratively sample candidate controllers and estimate their failure boundary region with the corresponding probability of failure in the specified ODD. The estimated probability of failure is subsequently used as a heuristic for the sampling distribution update. We show the effectiveness of this approach on multirotor Unmanned Aerial Vehicle (UAV) control system optimization. A dedicated Nonlinear Model Predictive Control (NMPC) scheme augmented by L 1 -adaptive control loop is developed for the quasi-physical model of a UAV and optimized to satisfy the safety requirements imposed by the ODD.

Published
2024 (in print)
Conference
AIAA AVIATION Forum, Las Vegas, US
Place
Las Vegas, US
BibTeX 
@INPROCEEDINGS{FITPUB13186,
   author = "Ji\v{r}\'{i} Nov\'{a}k and Ji\v{r}\'{i} Han\'{a}k and Peter Chud\'{y}",
   title = "Reliability-Based Control System Optimization in Uncertain Conditions",
   year = 2024,
   location = "Las Vegas, US",
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/13186"
}
Back to top