Real time defect detection in welds by ultrasonic means

A computer controlled weld quality assurance system has been developed to detect weld defects ultrasonically whilst welding is in progress. This system, including a flash analogue to digital converter and built-in memories to store sampled data, a peak characters extractor and a welding process controller, enabled welding processes to be controlled automatically and welding defects to be detected concurrently with welding. In this way, the weld quality could be satisfactorily assured if no defect was detected and the welding cost was minimised either through avoiding similar defects to occur or by stopping the welding process if repair was necessary. This work demonstrated that the high temperature field around the weld pool was the major source of difficulties and unreliabilities in defect detection during welding and, had to be taken into account in welding control by ultrasonic means. The high temperatures not only influence ultrasonic characteristic parameters which are the defect judgement and assessment criterion, but also introduce noise into signals. The signal averaging technique and statistical analysis based on B-scan data have proved their feasibility to increase 'signal to noise ratio' effectively and to judge or assess weld defects. The hardware and the software for the system is explained in this work. By using this system, real-time 'A-scan' signals on screen display, and, A-scan, B-scan or three dimensional results can be printed on paper, or stored on disks, and, as a result, weld quality could be fully computerized.

Grantor dc:publisher

Brunel University School of Engineering and Design PhD Theses

Year dc:date.issued

1992

Author dc:creator

• Lu, Yicheng

Advisor dc:contributor.advisor

• Fenn, R

• Quality assurance system
• Weld defects
• Ultrasonic detection
• Computer system

Language dc:language.iso

en

Repository record dc:identifier.uri

http://bura.brunel.ac.uk/handle/2438/4966

OAI identifier oai:identifier

oai:bura.brunel.ac.uk:2438/4966