Fault Detection and Fault Isolation play a major role in automated systems. In this context, emerging techniques, known as Active Fault Diagnosis, are gaining popularity. Active Fault Diagnosis strategies involve designing a minimally intrusive auxiliary input, in addition to the primary control input, which aims to generate more informative outputs for diagnosis. This work investigates a Safe Reinforcement Learning approach, Constrained Policy Optimization, to learn an effective policy for simultaneously performing Active Fault Diagnosis and trajectory tracking when the system experiences faults in the actuators.
Fault Detection e Fault Isolation hanno un ruolo di primo piano nei sistemi automatizzati. In questo ambito, delle tecniche emergenti, note come Active Fault Diagnosis, stanno acquistando popolarità. Le strategie di Active Fault Diagnosis prevedono il design di un input ausiliario minimamente intrusivo, oltre all’input di controllo primario, che ha lo scopo di generare output più informativi per la diagnosi. Questo lavoro investiga un approccio di Safe Reinforcement Learning, Constrained Policy Optimization, per apprendere una policy efficace per eseguire simultaneamente Active Fault Diagnosis e tracking di una traiettoria quando il sistema è soggetto a guasti negli attuatori.
A Safe Reinforcement Learning Approach for Active Fault Diagnosis
ZACCARIA, VALENTINA
2022/2023
Abstract
Fault Detection and Fault Isolation play a major role in automated systems. In this context, emerging techniques, known as Active Fault Diagnosis, are gaining popularity. Active Fault Diagnosis strategies involve designing a minimally intrusive auxiliary input, in addition to the primary control input, which aims to generate more informative outputs for diagnosis. This work investigates a Safe Reinforcement Learning approach, Constrained Policy Optimization, to learn an effective policy for simultaneously performing Active Fault Diagnosis and trajectory tracking when the system experiences faults in the actuators.File | Dimensione | Formato | |
---|---|---|---|
Zaccaria_Valentina.pdf
accesso riservato
Dimensione
3.44 MB
Formato
Adobe PDF
|
3.44 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/50914