Model uncertainty is not always taken into account in the engineering fields, notwithstanding it plays a key role in practical applications. This topic has led to the creation of specific kinds of literature, such as robust control or robust fault diagnosis. In the beginning, the uncertainty was quantified by the engineer. Instead, in the last decades, the system identification community has developed a branch of literature, called Robust identification, which studies different selection methods. The results of the uncertainty quantification and Robust identification procedures are directly related to the choices made by the engineer. The wrong choices could lead to a useless robust controller or robust fault diagnosis algorithm. This research proposes an automatic way to model the uncertainty information by employing the innovative kernel-based system identification. The resulting data-driven uncertainty model is then applied to the robust control design. Specifically, the mixed-sensitivity loop-shaping procedure is adopted to develop a robust controller for Single Input Single Output(SISO) and multi-model dynamic systems. The effectiveness of the proposed methodologies is proven by testing them on a benchmark problem (SISO system) and on a real application (multi-model system). Furthermore, the proposed data-driven uncertainty modeling is employed to design an algorithm for a robust model-based fault diagnosis problem. The proposed technique is able to reduce false alarms and shows good fault detectability. The other two contributions deal with practical applications. The first relies on a model-based fault diagnosis system applied to an Electro-Mechanical Actuator employed in a sliding gate system. Instead, the second is devoted to study a signal-based fault diagnosis system to detect and isolate some faults of a complex rotating machine.
(2024). Data-driven robust control and diagnosis : Theory and Application . Retrieved from https://hdl.handle.net/10446/283409 Retrieved from http://dx.doi.org/10.13122/978-88-97413-97-4
Data-driven robust control and diagnosis : Theory and Application
Valceschini, Nicholas
2024-01-01
Abstract
Model uncertainty is not always taken into account in the engineering fields, notwithstanding it plays a key role in practical applications. This topic has led to the creation of specific kinds of literature, such as robust control or robust fault diagnosis. In the beginning, the uncertainty was quantified by the engineer. Instead, in the last decades, the system identification community has developed a branch of literature, called Robust identification, which studies different selection methods. The results of the uncertainty quantification and Robust identification procedures are directly related to the choices made by the engineer. The wrong choices could lead to a useless robust controller or robust fault diagnosis algorithm. This research proposes an automatic way to model the uncertainty information by employing the innovative kernel-based system identification. The resulting data-driven uncertainty model is then applied to the robust control design. Specifically, the mixed-sensitivity loop-shaping procedure is adopted to develop a robust controller for Single Input Single Output(SISO) and multi-model dynamic systems. The effectiveness of the proposed methodologies is proven by testing them on a benchmark problem (SISO system) and on a real application (multi-model system). Furthermore, the proposed data-driven uncertainty modeling is employed to design an algorithm for a robust model-based fault diagnosis problem. The proposed technique is able to reduce false alarms and shows good fault detectability. The other two contributions deal with practical applications. The first relies on a model-based fault diagnosis system applied to an Electro-Mechanical Actuator employed in a sliding gate system. Instead, the second is devoted to study a signal-based fault diagnosis system to detect and isolate some faults of a complex rotating machine.File | Dimensione del file | Formato | |
---|---|---|---|
CollanaSAFD_Volume72_2024.pdf
accesso aperto
Versione:
publisher's version - versione editoriale
Licenza:
Creative commons
Dimensione del file
9.7 MB
Formato
Adobe PDF
|
9.7 MB | Adobe PDF | Visualizza/Apri |
Pubblicazioni consigliate
Aisberg ©2008 Servizi bibliotecari, Università degli studi di Bergamo | Terms of use/Condizioni di utilizzo