This thesis is based on the research of a method for the detection of anomalies in Swegon Operations S.r.l. heat pump units. Two machine learning techniques, namely Principal Component Analysis (PCA) and clustering, were studied for the research of such a method. Much of the research work was devoted to the collection and management of data, produced by the heat pump units. Data is crucial to carry forward the development of machine learning models. Finally, it is important to point out that throughout the work, the domain knowledge of the company was leveraged, which played a very important role in several aspects, such as to better understand the data we had in our hands.
Questa tesi si basa sulla ricerca di un metodo per il rilevamento di anomalie nelle unità a pompa di calore di Swegon Operations S.r.l.. Per la ricerca di tale metodo sono state studiate due tecniche di machine learning, ovvero la Principal Componet Analysis (PCA) e il clustering. Gran parte del lavoro di ricerca è stato dedicato alla raccolta e alla gestione dei dati prodotti dalle unità a pompa di calore. Essi sono fondamentali per portare avanti lo sviluppo dei modelli di machine learning. Infine, è importante sottolineare che durante tutto il lavoro è stata sfruttata la conoscenza del dominio dell'azienda, che ha giocato un ruolo molto importante in diversi aspetti, come ad esempio per comprendere meglio i dati che avevamo tra le mani.
Computational Approaches for Anomaly Detection in Heat Pump Units
URSINO, ALBERTO
2023/2024
Abstract
This thesis is based on the research of a method for the detection of anomalies in Swegon Operations S.r.l. heat pump units. Two machine learning techniques, namely Principal Component Analysis (PCA) and clustering, were studied for the research of such a method. Much of the research work was devoted to the collection and management of data, produced by the heat pump units. Data is crucial to carry forward the development of machine learning models. Finally, it is important to point out that throughout the work, the domain knowledge of the company was leveraged, which played a very important role in several aspects, such as to better understand the data we had in our hands.File | Dimensione | Formato | |
---|---|---|---|
Ursino_Alberto.pdf
accesso aperto
Dimensione
11.15 MB
Formato
Adobe PDF
|
11.15 MB | Adobe PDF | Visualizza/Apri |
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/20.500.12608/65952