In this study, professional algorithms based on decision trees and neural networks were chosen to address the problem of predictive maintenance. The methodological line followed started from a theoretical introduction on the concepts of maintenance and predictive maintenance and their connection with Artificial Intelligence. Subsequently, the characteristics of the data set used during the practical phase were explained. Have been employed some of the most innovative and reliable Python libraries, in order to carry out a procedure of pre-processing, processing and, finally, testing of the models. The results that emerged during the design phase were subjected to an in-depth critical analysis, aimed to highlight the advantages and disadvantages of the techniques used. In the light of the considerations emerged, possible improvements to the previously written code were proposed. The final improvement ideas concern more advanced and complex techniques, thanks to which the scalability of the model to further areas of operation can be made possible.
In questo studio sono stati scelti degli algoritmi professionali, basati su alberi decisionali e reti neurali, al fine di affrontare il problema della manutenzione predittiva. La linea metodologica seguita è partita da un’introduzione teorica sui concetti di manutenzione e manutenzione predittiva e sul loro nesso con l’Intelligenza Artificiale. Successivamente, sono state esposte le caratteristiche del dataset utilizzato durante la fase pratica. Sono state impiegate alcune delle più innovative e affidabili librerie di Python, al fine di effettuare una procedura di pre-processing, di elaborazione e, infine, di testing dei modelli. I risultati emersi durante la fase progettuale sono stati sottoposti a un’approfondita analisi critica, volta a evidenziare i vantaggi e gli svantaggi delle tecniche adoperate. Alla luce delle considerazioni emerse, sono stati proposti dei possibili miglioramenti del codice scritto precedentemente. Le idee migliorative finali riguardano tecniche più avanzate e complesse, grazie alle quali può rendersi possibile la scalabilità del modello a ulteriori settori operativi.
Manutenzione predittiva applicata all’ambito industriale
POPA, VLADUT GABRIEL
2023/2024
Abstract
In this study, professional algorithms based on decision trees and neural networks were chosen to address the problem of predictive maintenance. The methodological line followed started from a theoretical introduction on the concepts of maintenance and predictive maintenance and their connection with Artificial Intelligence. Subsequently, the characteristics of the data set used during the practical phase were explained. Have been employed some of the most innovative and reliable Python libraries, in order to carry out a procedure of pre-processing, processing and, finally, testing of the models. The results that emerged during the design phase were subjected to an in-depth critical analysis, aimed to highlight the advantages and disadvantages of the techniques used. In the light of the considerations emerged, possible improvements to the previously written code were proposed. The final improvement ideas concern more advanced and complex techniques, thanks to which the scalability of the model to further areas of operation can be made possible.File | Dimensione | Formato | |
---|---|---|---|
Tesi Popa (1).pdf
accesso aperto
Dimensione
4.72 MB
Formato
Adobe PDF
|
4.72 MB | Adobe PDF | Visualizza/Apri |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.12075/21114