Authors: Huanwei Yu Zhen Ye Shanben Chen
Publish Date: 2013/03/23
Volume: 68, Issue: 9-12, Pages: 2713-2727
Abstract
In this paper an intelligent fuzzy system based on arc spectral information was proposed to recognize the weld penetration status A spectrometer and a visual sensor were used simultaneously to collect the arc spectral and visual information about the Al–Mg alloy pulsed GTAW process respectively Principal component analysis was utilized for spectral line identification redundancy removal and spectral characteristic signals extraction wavelet packet denoising technique was used to filter these signals The relationship between the penetration status under different welding conditions and the corresponding spectral signals was discussed and clarified Finally a fuzzy system based on the spectral signals was developed and successfully used to estimate the percentage index of the weld penetration status With a binary decision of penetration index the average recognition rate for the penetration status about the specific welding conditions was no less than 821
Keywords: