This article proposes the use of an expert knowledge-based system capable of identifying possible machine tool failures caused by accidental events (e.g., cable disconnection, incorrect parameterization of machining, impact event, etc.). The proposed approach aims to identify the unpredictable causes of failures, starting from the analysis of the process data provided by the PLC Data Logger, without requiring to sensorize the machine in order to collect ad hoc condition monitoring data. To this end, it uses the experts rules and Fuzzy Logic algorithms to activate data analysis based on known machine fault conditions. The proposed approach has been validated on real case studies. A prototype system was developed in Python to identify electrospindle failures that occur when the spindle of a CNC machining center for woodworking is subjected to a strong axial impact. The results show that the proposed system is capable of effectively detecting failures caused by impact events and reduces the time needed for the diagnosis by 80%.
A Fuzzy Knowledge-Based System for Diagnosing Unpredictable Failures in CNC Machine Tools
Antony Colasante;Silvia Ceccacci;
2019-01-01
Abstract
This article proposes the use of an expert knowledge-based system capable of identifying possible machine tool failures caused by accidental events (e.g., cable disconnection, incorrect parameterization of machining, impact event, etc.). The proposed approach aims to identify the unpredictable causes of failures, starting from the analysis of the process data provided by the PLC Data Logger, without requiring to sensorize the machine in order to collect ad hoc condition monitoring data. To this end, it uses the experts rules and Fuzzy Logic algorithms to activate data analysis based on known machine fault conditions. The proposed approach has been validated on real case studies. A prototype system was developed in Python to identify electrospindle failures that occur when the spindle of a CNC machining center for woodworking is subjected to a strong axial impact. The results show that the proposed system is capable of effectively detecting failures caused by impact events and reduces the time needed for the diagnosis by 80%.File | Dimensione | Formato | |
---|---|---|---|
1-s2.0-S2351978920301220-main.pdf
accesso aperto
Tipologia:
Documento in post-print (versione successiva alla peer review e accettata per la pubblicazione)
Licenza:
Copyright dell'editore
Dimensione
677.36 kB
Formato
Adobe PDF
|
677.36 kB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.