A study on using scanning acoustic microscopy and neural network techniques to evaluate the quality of resistance spot welding

Abstract

This paper shows that the quality of resistance spot welds can be evaluated using scanning acoustic microscopy (SAM). Two-layered coated spot-welded samples are investigated utilising a wide-field short-pulse scanning acoustic microscope with operation frequencies of 25, 50 and 100 MHz. Geometrical parameters, e.g. nugget area, maximum axis of nugget, and minimum axis of nugget, are acquired from C-scan images of weld nuggets using mathematical morphology techniques. These parameters serve as inputs for an artificial neural network (ANN) model to evaluate the quality of spot welds. The output of the model during the training process comprises the results of nugget peeling tests and expert opinions. The ANN can provide suggestions on weld quality with a higher than 95% correctness. A JAVA computer program is developed for image processing, ANN training, and ANN testing. With this model, the computer program can render the quality of spot welds that are close to those achieved using off-line destructive method.