Dynamic estimation of joint penetration by deep learning from weld pool image

Yongchao Cheng; Shujun Chen; Jun Xiao; Yu Ming Zhang

doi:10.1080/13621718.2021.1896141

Dynamic estimation of joint penetration by deep learning from weld pool image

Yongchao Cheng, Shujun Chen, Jun Xiao, Yu Ming Zhang

Research output: Contribution to journal › Article › peer-review

30 Scopus citations

Abstract

This work aims at a novel approach to estimate the root-pass penetration towards its feedback control, in which the real penetration is measured by the backside bead width. The major challenge is that it happens under the workpiece and likely cannot be directly observable. The dynamic evolution of the weld pool surface has been analysed to design an active vision method monitoring the pool surface, yet fundamentally correlated to the unobservable penetration. The designed convolutional neural network model is trained, validated, and tested for recognising the weld penetration with satisfactory accuracy.

Original language	English
Pages (from-to)	279-285
Number of pages	7
Journal	Science and Technology of Welding and Joining
Volume	26
Issue number	4
DOIs	https://doi.org/10.1080/13621718.2021.1896141
State	Published - 2021

Bibliographical note

Publisher Copyright:
© 2021 Institute of Materials, Minerals and Mining. Published by Taylor & Francis on behalf of the Institute.

Funding

This work was supported by Beijing Municipal Natural Science Foundation: [Grant Number 3192004]; National Natural Science Foundation of China: [Grant Number 51975014], and General Program of Science and Technology Development Project of Beijing Municipal Education Commission: [Grant Number KM201910005034]. This work was supported by the International Research Cooperation Seed Fund of Beijing University of Technology; National Natural Science Foundation of China [grant numbers 51775007, 51975014]; Beijing Municipal Natural Science Foundation [grant number 3192004].

Funders	Funder number
National Natural Science Foundation of China (NSFC)	51975014
Beijing Municipal Commission of Education	KM201910005034
Beijing University of Technology	51775007
Beijing Municipal Natural Science Foundation	3192004

Keywords

Active vision
CNN
penetration estimation
weld pool surface

ASJC Scopus subject areas

General Materials Science
Condensed Matter Physics

Access to Document

10.1080/13621718.2021.1896141

Cite this

@article{f629d771292b4020aee94a5bcf735643,

title = "Dynamic estimation of joint penetration by deep learning from weld pool image",

abstract = "This work aims at a novel approach to estimate the root-pass penetration towards its feedback control, in which the real penetration is measured by the backside bead width. The major challenge is that it happens under the workpiece and likely cannot be directly observable. The dynamic evolution of the weld pool surface has been analysed to design an active vision method monitoring the pool surface, yet fundamentally correlated to the unobservable penetration. The designed convolutional neural network model is trained, validated, and tested for recognising the weld penetration with satisfactory accuracy.",

keywords = "Active vision, CNN, penetration estimation, weld pool surface",

author = "Yongchao Cheng and Shujun Chen and Jun Xiao and Zhang, \{Yu Ming\}",

note = "Publisher Copyright: {\textcopyright} 2021 Institute of Materials, Minerals and Mining. Published by Taylor \& Francis on behalf of the Institute.",

year = "2021",

doi = "10.1080/13621718.2021.1896141",

language = "English",

volume = "26",

pages = "279--285",

number = "4",

}

TY - JOUR

T1 - Dynamic estimation of joint penetration by deep learning from weld pool image

AU - Cheng, Yongchao

AU - Chen, Shujun

AU - Xiao, Jun

AU - Zhang, Yu Ming

PY - 2021

Y1 - 2021

N2 - This work aims at a novel approach to estimate the root-pass penetration towards its feedback control, in which the real penetration is measured by the backside bead width. The major challenge is that it happens under the workpiece and likely cannot be directly observable. The dynamic evolution of the weld pool surface has been analysed to design an active vision method monitoring the pool surface, yet fundamentally correlated to the unobservable penetration. The designed convolutional neural network model is trained, validated, and tested for recognising the weld penetration with satisfactory accuracy.

AB - This work aims at a novel approach to estimate the root-pass penetration towards its feedback control, in which the real penetration is measured by the backside bead width. The major challenge is that it happens under the workpiece and likely cannot be directly observable. The dynamic evolution of the weld pool surface has been analysed to design an active vision method monitoring the pool surface, yet fundamentally correlated to the unobservable penetration. The designed convolutional neural network model is trained, validated, and tested for recognising the weld penetration with satisfactory accuracy.

KW - Active vision

KW - CNN

KW - penetration estimation

KW - weld pool surface

UR - http://www.scopus.com/inward/record.url?scp=85102955566&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85102955566&partnerID=8YFLogxK

U2 - 10.1080/13621718.2021.1896141

DO - 10.1080/13621718.2021.1896141

M3 - Article

AN - SCOPUS:85102955566

SN - 1362-1718

VL - 26

SP - 279

EP - 285

JO - Science and Technology of Welding and Joining

JF - Science and Technology of Welding and Joining

IS - 4

ER -

Dynamic estimation of joint penetration by deep learning from weld pool image

Abstract

Bibliographical note

Funding

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this