Use of deep neural networks for clogging detection in the Submerged Entry Nozzle of the continuous casting

The continuous casting process, used in the manufacture of steel plates, is currently the most economical and efficient way of production. Although continuous casting is a well-established process, many associated issues remain to be resolved, including obstructions that occur in the Submerged Entry...

Full description

Saved in:
Bibliographic Details
Published in:Expert systems with applications 2024-03, Vol.238, p.121963, Article 121963
Main Authors: Diniz, Ana Paula Miranda, Ciarelli, Patrick Marques, Salles, Evandro Ottoni Teatini, Coco, Klaus Fabian
Format: Article
Language:English
Subjects:
Clogging
Continuous casting
Deep neural networks
Machine learning
Submerged Entry Nozzle (SEN)
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The continuous casting process, used in the manufacture of steel plates, is currently the most economical and efficient way of production. Although continuous casting is a well-established process, many associated issues remain to be resolved, including obstructions that occur in the Submerged Entry Nozzle (SEN) that controls the flow of steel between the tundish and the mold. Clogging on SEN not only impairs the quality of the product but also results in lower process yield, resulting in losses. In this context, we evaluated three methodologies based on deep neural networks to detect the occurrences of clogging in a steel industry from historical data of process variables. As a baseline, models based on traditional artificial neural networks were used. The overall performance of the classifiers developed here showed very promising results in real data with unbalanced classes. In particular, the method employing spatiotemporal analysis obtained a remarkably superior performance with respect to the other methods, achieving a Recall and a Precision of almost 99% and 98%, respectively. The code is available for public access through the link: https://github.com/apaulinhamd/Deep-Neural-Networks-For-Clogging-Detection. •Use of exclusively data-oriented models to identify clogging occurrences.•Application of data pre-processing techniques without loss of information.•Use of clogging classifiers with the lowest false negative and false positive rates.•Classification in sufficient time for the adoption of preventive actions.
ISSN:0957-4174
1873-6793
DOI:10.1016/j.eswa.2023.121963