STEEL SAMPLES WELDING DEFECTS DETECTION BY ESTIMATING ENERGY OF CLUSTERS OF ACOUSTIC EMISSION SIGNALS

Abstract

Acoustic emission (AE) detection of 5005005 mm size St3 steel samples welding defects. The informative parameters of AE signals recorded during welding with inserts of titanium and aluminum inserted into the root of the weld are analyzed. AE-signals recording was performed by welding diagnostic unit with automatically changing “floating” selection threshold. For coordinates of defects estimation clustering was performed by digital form and AE signals parameters. A comparative analysis of the clustering results was performed for each cluster. In this case, the total count of AE signals, their energy, and threshold levels were set, which were determined in accordance with the proposed algorithm. The degree of danger of defects in the weld was monitored by exceeding AE signal energy threshold level. If the AE signal had a second energy peak and when it exceeds the energy selection thresholds, the weld was rejected. It was experimentally shown that the titanium insert was a dangerous defect when the threshold energy level was exceeded, which was detected both with a high total count of AE signals and with a small number of recorded signals. In accordance with the proposed method, when introducing an aluminum insert into the weld, the defect was not considered dangerous.