Friction stir welding is a solid-state welding process. The technology is used in high-precision applications such asaerospace. Thus, monitoring the weld quality is highly relevant for detecting inaccurate welds. Various studies have showna significant dependence of the weld quality on the welding speed and the rotational speed of the tool. Frequently, anunsuitable setting of these parameters can be detected by visually examining the resulting surface defects, such as increasedflash formation or surface galling. The visual inspection for these defects is often conducted by humans and is thereforeassociated with increased costs and personnel allocation.In this work, a deep learning approach to automatically detect irregularities on the weld surface is introduced. A total of112 welds with a total length of 18.4 metres were made to train and test of the artificial neural networks. Colour images ofthe welds were made using a digital camera, while images of the weld surface topography were made with a threedimensionalprofilometer. The approach consisted of a two-step procedure. First, an object detector using a neural networklocalised the friction stir weld on the image. Second, a neural network classified the surface properties of the weld seam.The object detector localised the friction stir welds with an Intersection over Union up to 89.5%. The best result inclassifying the surface properties was achieved by using the topography images. Here, a classification accuracy of 92.1%was reached by the DenseNet-121 convolutional neural network. The results will form the basis for the future developmentof a parameter optimization method for friction stir welding.
展开▼
