This paper presents a novel machine learning-based prognostic approach for on-board electromechanical actuators. The study is centered around overcoming the limitations of model-based prognostic frameworks that rely on expensive optimization processes. Machine learning techniques were employed to map system signal characteristics directly into parameters related to fault simulation. A first test, utilizing only five of eight implemented fault types, demonstrates a highly promising potential of artificial neural networks to predict and detect faults with minimal error. A second test expands the investigation to include all fault types and provides an analysis of the model’s robustness, error rates, and computational costs. The practical outcome of the work is a viable real-time solution for fault detection and characterization in electromechanical actuators, highlighting the efficiency and effectiveness of machine learning techniques for industrial applications.
Machine Learning Based Prognostics of On-Board Electromechanical Actuators / Minisci, Edmondo; Dalla Vedova, Matteo; Alimhillaj, Parid; Baldo, Leonardo; Maggiore, Paolo (LECTURE NOTES ON MULTIDISCIPLINARY INDUSTRIAL ENGINEERING). - In: Lecture Notes on Multidisciplinary Industrial EngineeringELETTRONICO. - [s.l] : Springer Nature, 2024. - ISBN 9783031489327. - pp. 148-159 [10.1007/978-3-031-48933-4_15]
Machine Learning Based Prognostics of On-Board Electromechanical Actuators
Minisci, Edmondo;Dalla Vedova, Matteo;Baldo, Leonardo;Maggiore, Paolo
2024
Abstract
This paper presents a novel machine learning-based prognostic approach for on-board electromechanical actuators. The study is centered around overcoming the limitations of model-based prognostic frameworks that rely on expensive optimization processes. Machine learning techniques were employed to map system signal characteristics directly into parameters related to fault simulation. A first test, utilizing only five of eight implemented fault types, demonstrates a highly promising potential of artificial neural networks to predict and detect faults with minimal error. A second test expands the investigation to include all fault types and provides an analysis of the model’s robustness, error rates, and computational costs. The practical outcome of the work is a viable real-time solution for fault detection and characterization in electromechanical actuators, highlighting the efficiency and effectiveness of machine learning techniques for industrial applications.Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/11583/2991506
Attenzione
Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo