Prediction of Bead Geometry Using a Two-Stage SVM–ANN Algorithm for Automated Tungsten Inert Gas (TIG) Welds

Rohit Kshirsagar, Steve Jones, Jonathan Lawrence, Jim Tabor

Research output: Contribution to journalArticle

Abstract

Prediction of weld bead geometry is critical for any welding process, since several mechanical properties of the weldment depend on this. Researchers have used artificial neural networks (ANNs) to predict the bead geometry based on the input parameters for a welding process; however, the number of hidden layers used in these ANNs are limited to one due to the small amount of data usually available through experiments. This results in a reduction in the accuracy of prediction. Such ANNs are also incapable of capturing sudden changes in the input–output trends; for example, where a wide range of heat inputs results in flat crown (zero crown height), but any further reduction in the current sharply increases the crown height. In this study, it was found that above mentioned issues can be resolved on using a two-stage algorithm consisting of support vector machine (SVM) and an ANN. The two-stage SVM–ANN algorithm significantly improved the accuracy of prediction and could be used as a replacement for the multiple hidden layer ANN, without requiring additional data for training. The improvement in prediction was evident near regions of sudden changes in the input–output correlation and can lead to a better prediction of mechanical properties.
Original languageEnglish
Article number39
Number of pages18
JournalJournal of Manufacturing and Materials Processing
Volume3
Issue number2
DOIs
Publication statusPublished - 8 May 2019

Fingerprint

Inert gases
Tungsten
Welds
Neural networks
Geometry
Welding
Mechanical properties
Support vector machines
Experiments

Bibliographical note

This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/)

Keywords

  • bead geometry prediction
  • support vector machines
  • artificial neural networks
  • data classification

Cite this

Prediction of Bead Geometry Using a Two-Stage SVM–ANN Algorithm for Automated Tungsten Inert Gas (TIG) Welds. / Kshirsagar, Rohit; Jones, Steve; Lawrence, Jonathan; Tabor, Jim.

In: Journal of Manufacturing and Materials Processing, Vol. 3, No. 2, 39, 08.05.2019.

Research output: Contribution to journalArticle

@article{d4a2808fa6ee4ab4a213638b9048b0d3,
title = "Prediction of Bead Geometry Using a Two-Stage SVM–ANN Algorithm for Automated Tungsten Inert Gas (TIG) Welds",
abstract = "Prediction of weld bead geometry is critical for any welding process, since several mechanical properties of the weldment depend on this. Researchers have used artificial neural networks (ANNs) to predict the bead geometry based on the input parameters for a welding process; however, the number of hidden layers used in these ANNs are limited to one due to the small amount of data usually available through experiments. This results in a reduction in the accuracy of prediction. Such ANNs are also incapable of capturing sudden changes in the input–output trends; for example, where a wide range of heat inputs results in flat crown (zero crown height), but any further reduction in the current sharply increases the crown height. In this study, it was found that above mentioned issues can be resolved on using a two-stage algorithm consisting of support vector machine (SVM) and an ANN. The two-stage SVM–ANN algorithm significantly improved the accuracy of prediction and could be used as a replacement for the multiple hidden layer ANN, without requiring additional data for training. The improvement in prediction was evident near regions of sudden changes in the input–output correlation and can lead to a better prediction of mechanical properties.",
keywords = "bead geometry prediction, support vector machines, artificial neural networks, data classification",
author = "Rohit Kshirsagar and Steve Jones and Jonathan Lawrence and Jim Tabor",
note = "This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/)",
year = "2019",
month = "5",
day = "8",
doi = "10.3390/jmmp3020039",
language = "English",
volume = "3",
journal = "Journal of Manufacturing and Materials Processing",
issn = "2504-4494",
number = "2",

}

TY - JOUR

T1 - Prediction of Bead Geometry Using a Two-Stage SVM–ANN Algorithm for Automated Tungsten Inert Gas (TIG) Welds

AU - Kshirsagar, Rohit

AU - Jones, Steve

AU - Lawrence, Jonathan

AU - Tabor, Jim

N1 - This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/)

PY - 2019/5/8

Y1 - 2019/5/8

N2 - Prediction of weld bead geometry is critical for any welding process, since several mechanical properties of the weldment depend on this. Researchers have used artificial neural networks (ANNs) to predict the bead geometry based on the input parameters for a welding process; however, the number of hidden layers used in these ANNs are limited to one due to the small amount of data usually available through experiments. This results in a reduction in the accuracy of prediction. Such ANNs are also incapable of capturing sudden changes in the input–output trends; for example, where a wide range of heat inputs results in flat crown (zero crown height), but any further reduction in the current sharply increases the crown height. In this study, it was found that above mentioned issues can be resolved on using a two-stage algorithm consisting of support vector machine (SVM) and an ANN. The two-stage SVM–ANN algorithm significantly improved the accuracy of prediction and could be used as a replacement for the multiple hidden layer ANN, without requiring additional data for training. The improvement in prediction was evident near regions of sudden changes in the input–output correlation and can lead to a better prediction of mechanical properties.

AB - Prediction of weld bead geometry is critical for any welding process, since several mechanical properties of the weldment depend on this. Researchers have used artificial neural networks (ANNs) to predict the bead geometry based on the input parameters for a welding process; however, the number of hidden layers used in these ANNs are limited to one due to the small amount of data usually available through experiments. This results in a reduction in the accuracy of prediction. Such ANNs are also incapable of capturing sudden changes in the input–output trends; for example, where a wide range of heat inputs results in flat crown (zero crown height), but any further reduction in the current sharply increases the crown height. In this study, it was found that above mentioned issues can be resolved on using a two-stage algorithm consisting of support vector machine (SVM) and an ANN. The two-stage SVM–ANN algorithm significantly improved the accuracy of prediction and could be used as a replacement for the multiple hidden layer ANN, without requiring additional data for training. The improvement in prediction was evident near regions of sudden changes in the input–output correlation and can lead to a better prediction of mechanical properties.

KW - bead geometry prediction

KW - support vector machines

KW - artificial neural networks

KW - data classification

UR - http://www.mendeley.com/research/prediction-bead-geometry-using-twostage-svmann-algorithm-automated-tungsten-inert-gas-tig-welds

U2 - 10.3390/jmmp3020039

DO - 10.3390/jmmp3020039

M3 - Article

VL - 3

JO - Journal of Manufacturing and Materials Processing

JF - Journal of Manufacturing and Materials Processing

SN - 2504-4494

IS - 2

M1 - 39

ER -