<p>With the wide application of lightweight materials, traditional welding methods are not fully applicable in some occasions. Self-piercing rivet (SPR), a method used for joining sheet materials, is of increasing interest in automobile industries due to its suitability for joining lightweight, high strength and dissimilar materials. During the machining process, it is very easy to form cracks on the surface of rivets for many reasons. And in terms of corrosion properties, fatigue properties and many other properties, the cracks on the SPR joint surface might have a significant adverse impact. For these reasons, the SPR joint button needs to be checked and classified to meet the required standards. The traditional crack inspection is manual detection which is time-consuming and need skilled engineers to distinguish features by eyes. It is too subjective and has great limitations. On the basis of back propagation neural network (BP) and image processing, this paper proposed a crack detection and evaluation method which applied the idea of ensemble learning to the detection algorithm for SPR button images, and compared the algorithm with five other methods using a feature extraction separately at the same time. First of all, in order to reduce the influence of different background conditions, the crack detectors based on local area features is applied. Sub-images are cropped from the button images and preprocessed into three categories (i.e., cracks, edges and smooth regions) as training samples by three image preprocessing methods. In sub-images level, 5 neural networks are trained with the input of different extracted features, respectively. Secondly, to overcome the representation limitation of one single extracted feature, a weighted combination of 5 neural networks using ensemble learning idea is developed, and there will be a parameters optimization to find the best parameters combination. And then, a search algorithm is developed to extend the application of the learned model from sub-images into the original button images. The area where the crack was detected will be marked with a green rectangle. Finally, an evaluation system based on the characteristics of SPR button images is proposed to compare and analysis these different application results. The preliminary results on non-cracked and cracked button images show that the proposed crack detection method of weighted combination is an effective approach. And the methods of weighted combination which in different targets both get better crack detection results compared to other crack detection models with a single extracted feature of input.</p>

Crack Detection and Evaluation Method for Self-Piercing Riveting Button Images Based on BP Neural Network

Paper Proceedings: https://asmedigitalcollection.asme.org/IMECE/IMECE2021/volume/85543

ASME 2021 International Mechanical Engineering Congress and Exposition (IMECE2021)

Supporting the need among researchers worldwide for additional online and multimedia content, ASME now offers the ASME Conference Video Collection. This platform will collect and provide long-term access to conference presentation videos, encompassing all technology areas within ASME's technical conference and events program. About 30,000-40,000 engineers across all mechanical engineering and related disciplines participate in ASME conferences each year. This includes authors of approximately 6,000 to 8,000 technical papers published by ASME, the majority of whom also submit video presentations for our virtual conferences. The video collection will be available as an add-on to existing product packages or as a separate subscription.