In vapour compression refrigeration systems, oil circulates to lubricate moving parts. However, due to its low miscibility with most environmentally friendly refrigerants, such as ammonia, it is retained in some parts of the system causing losses in the overall system efficiency. Therefore, this paper focusses on the investigation of the fault characteristics of oil-retention by simulating this fault using a test facility. Based on the obtained dataset, a data-driven fault diagnosis approach is derived. Furthermore, a genetic algorithm is used for the selection of characteristic features, which are finally defined as input parameters for an exemplary implemented classification algorithm. It is also demonstrated how this classification algorithm correctly distinguishes multiple system states from one another.
Van De Sand, R., Falk, C., Corasaniti, S., Reiff-Stephan, J. (2019). A data-driven fault diagnosis approach towards oil retention in vapour compression refrigeration systems. In CANDO-EPE 2019 - Proceedings: IEEE 2nd International Conference and Workshop in Obuda on Electrical and Power Engineering (pp.197-202). IEEE [10.1109/CANDO-EPE47959.2019.9111046].
A data-driven fault diagnosis approach towards oil retention in vapour compression refrigeration systems
Corasaniti S.
;
2019-01-01
Abstract
In vapour compression refrigeration systems, oil circulates to lubricate moving parts. However, due to its low miscibility with most environmentally friendly refrigerants, such as ammonia, it is retained in some parts of the system causing losses in the overall system efficiency. Therefore, this paper focusses on the investigation of the fault characteristics of oil-retention by simulating this fault using a test facility. Based on the obtained dataset, a data-driven fault diagnosis approach is derived. Furthermore, a genetic algorithm is used for the selection of characteristic features, which are finally defined as input parameters for an exemplary implemented classification algorithm. It is also demonstrated how this classification algorithm correctly distinguishes multiple system states from one another.File | Dimensione | Formato | |
---|---|---|---|
19 CANDO EPE.pdf
solo utenti autorizzati
Tipologia:
Versione Editoriale (PDF)
Licenza:
Copyright dell'editore
Dimensione
1.25 MB
Formato
Adobe PDF
|
1.25 MB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.