Maintenance records contain information about equipment degradation, which is typically not used as “data” for maintenance modeling and optimization given the difficulty of analyzing text to extract relevant and usable information. In this work, we present a method for using such textual information by: 1) the identification of equipment degradation states through clustering the maintenance records using a Convolutional Neural Network (CNN) and 2) the development of a stochastic multi-state degradation model based on the information therein. The method is applied to a database of maintenance records collected from excavator buckets used for mining. The results show that the proposed CNN-based clustering method identifies homogeneous clusters of maintenance records representing different equipment degradation states and that the proposed stochastic multi-state degradation model based on this information can be used for improving maintenance.
A novel method for maintenance record clustering and its application to a case study of maintenance optimization
Yang Z.;Baraldi P.;Zio E.
2020-01-01
Abstract
Maintenance records contain information about equipment degradation, which is typically not used as “data” for maintenance modeling and optimization given the difficulty of analyzing text to extract relevant and usable information. In this work, we present a method for using such textual information by: 1) the identification of equipment degradation states through clustering the maintenance records using a Convolutional Neural Network (CNN) and 2) the development of a stochastic multi-state degradation model based on the information therein. The method is applied to a database of maintenance records collected from excavator buckets used for mining. The results show that the proposed CNN-based clustering method identifies homogeneous clusters of maintenance records representing different equipment degradation states and that the proposed stochastic multi-state degradation model based on this information can be used for improving maintenance.File | Dimensione | Formato | |
---|---|---|---|
2020_RESS_Zhe_Baraldi_Zio_maintenance-record-clustering.pdf
Accesso riservato
:
Publisher’s version
Dimensione
1.04 MB
Formato
Adobe PDF
|
1.04 MB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.