In the production of steel strips, the fulfillment of required product properties is a key factor to improve the company’s productivity and competitiveness. Product characteristics can be evaluated online throughout the length of the strip by means of non–destructive tests such as the IMPOC whose output signal is related to mechanical properties and their uniformity. In this work, a novel approach based on the use of deep–neural–networks and advanced analytics is used to develop a model for the prediction of IMPOC signal from process parameters. The model provides plant managers with an insight into the relationships among process conditions, product characteristics and mechanical properties in order to suitably set up process parameters to meet product requirements. In this work, different model architectures and data processing techniques are evaluated leading an overall prediction error lower than 5% that puts the basis for their integration into the plant.

Prediction of steel coils mechanical properties and microstructure by using deep learning and advanced data preprocessing techniques

Vannucci M.
;
Colla V.;Mocci C.;
2021-01-01

Abstract

In the production of steel strips, the fulfillment of required product properties is a key factor to improve the company’s productivity and competitiveness. Product characteristics can be evaluated online throughout the length of the strip by means of non–destructive tests such as the IMPOC whose output signal is related to mechanical properties and their uniformity. In this work, a novel approach based on the use of deep–neural–networks and advanced analytics is used to develop a model for the prediction of IMPOC signal from process parameters. The model provides plant managers with an insight into the relationships among process conditions, product characteristics and mechanical properties in order to suitably set up process parameters to meet product requirements. In this work, different model architectures and data processing techniques are evaluated leading an overall prediction error lower than 5% that puts the basis for their integration into the plant.
File in questo prodotto:
File Dimensione Formato  
MAS2021_published.pdf

accesso aperto

Licenza: Creative commons (selezionare)
Dimensione 1.62 MB
Formato Adobe PDF
1.62 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/11382/551091
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
social impact