Detection in a Robotised Short Circuting GMA Welding using Neural Networks

Document type: Conference Papers
Peer reviewed: Yes
Author(s): Linus Pettersson, Stefan Adolfsson, Gunnar Bolmsjö, Ingvar Claesson
Title: Detection in a Robotised Short Circuting GMA Welding using Neural Networks
Conference name: IEEE International Symposium on Intelligent Signal Processing and Communication System
Year: 1999
City: Phuket, Thailand
Organization: Blekinge Institute of Technology
Department: Department of Telecommunications and Signal Processing (Institutionen för telekommunikation och signalbehandling)
Department of Telecommunications and Signal Processing S-372 25 Ronneby
+46 455 38 50 00
Authors e-mail: lar@bth.se
Language: English
Abstract: Today it is both time and cost consuming to check the quality of a weld when it is done off-line by an experienced weld operator. Therefor the need for an automatic detection is urgent in order to reduce production costs. There are systems for monitoring the quality of a weld commercially available but there is still some research that has to be done in this area in order to increase the reliability.
The method presented in this paper is neural network based and considers short circuiting GMA welding, but there is no obstacle for this solution to work on other types of robotised welding. By presenting the weld voltage to a neural network, the network is able to detect defects in the weld joint.
Testresults have shown that the detection rate is 100 percent and false alarms are nonexisting.
Subject: Signal Processing\Detection and Classification
Edit