Automatic Detection of Burn-through in GMA Welding Using a Parametric Model

Document type: Journal Articles
Article type: Original article
Peer reviewed: Yes
Author(s): Stefan Adolfsson, Klas Ericson, Anders Grennberg
Title: Automatic Detection of Burn-through in GMA Welding Using a Parametric Model
Journal: Mechanical Systems & Signal Processing
Year: 1996
Volume: 10
Issue: 5
Pagination: 633-651
ISSN: 0888-3270
Publisher: Academic Press
City: London
ISI number: A1996VP07600009
Organization: Blekinge Institute of Technology
Department: Dept. of Signal Processing (Institutionen för signalbehandling)
Dept. of Signal Processing S-372 25 Ronneby
+46 455 780 00
Authors e-mail:
Language: English
Abstract: This paper addresses the problem of automatic detection of burn-through in weld joints. Gas metal are (GMA) welding with pulsed current is used, and welding voltage and current are recorded. As short-circuitings are common between the welding electrode and the work piece during burn-through, a short-circuit detector is developed to detect these events. To detect another specific characteristic of burn-through-a broadband long-lasting voltage component-this detector is combined with a square-law detector. This second detector is based on a non-linear modification of an autoregressive model with extra input (ARX-model) of the welding process. The results obtained from this compound detector indicate that it is possible to detect burn-through in the welds automatically. The work also indicates that it is possible to design an on-line monitoring system for robotic GMA welding.
Subject: Signal Processing\Detection and Classification
Note: Published in issue October