One of the pillars of the smart factory concept within the Industry 4.0 paradigm is the capability to monitor the health conditions of production systems and their critical components in a continuous and effective way. This could be enabled through the implementation of innovative diagnosis, prognosis and predictive maintenance actions. A wide literature has been devoted to methodologies to monitor the manufacturing process and the tool wear. A parallel research field is dedicated to isolate the health condition of the machine tool from the production process and external source of noise. This study presents a novel solution for machine health condition monitoring based on the so-called “fingerprint” cycle approach. A fingerprint cycle is a pre-defined test cycle in no-load conditions, where the axes and the spindle are activated in a sequential order. Several signals are extracted from the machine controller to characterize the current health state of the machine. The method is suitable to separate drifts, trends and shifts in CNC signals caused by a change in machine tool health condition from any variation related to the cutting process and external factors. A machine learning method that combines Principal Component Analysis and statistical process monitoring allows one to quickly detect degraded conditions affecting one or multiple critical components. A real case study is presented to highlight the potentials and benefits provided by the proposed approach.

Fingerprint analysis for machine tool health condition monitoring

Garghetti F.;Grasso M.;Colosimo B. M.
2021-01-01

Abstract

One of the pillars of the smart factory concept within the Industry 4.0 paradigm is the capability to monitor the health conditions of production systems and their critical components in a continuous and effective way. This could be enabled through the implementation of innovative diagnosis, prognosis and predictive maintenance actions. A wide literature has been devoted to methodologies to monitor the manufacturing process and the tool wear. A parallel research field is dedicated to isolate the health condition of the machine tool from the production process and external source of noise. This study presents a novel solution for machine health condition monitoring based on the so-called “fingerprint” cycle approach. A fingerprint cycle is a pre-defined test cycle in no-load conditions, where the axes and the spindle are activated in a sequential order. Several signals are extracted from the machine controller to characterize the current health state of the machine. The method is suitable to separate drifts, trends and shifts in CNC signals caused by a change in machine tool health condition from any variation related to the cutting process and external factors. A machine learning method that combines Principal Component Analysis and statistical process monitoring allows one to quickly detect degraded conditions affecting one or multiple critical components. A real case study is presented to highlight the potentials and benefits provided by the proposed approach.
2021
Proceedings of the 17th IFAC Symposium on Information Control Problems in Manufacturing INCOM 2021
Control chart
Fingerprint
Health monitoring
Industry 4.0
Machine tool
Principal component analysis
File in questo prodotto:
File Dimensione Formato  
Fingerprint analysis for machine tool health condition monitoring.pdf

accesso aperto

: Publisher’s version
Dimensione 1.2 MB
Formato Adobe PDF
1.2 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11311/1192009
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 4
  • ???jsp.display-item.citation.isi??? 1
social impact