Abstract

In resistance spot welding, the quality of welds is not only affected by the correct design of the welding cycle, but also by the electrode degradation that occurs over time. This work proposes a novel approach to indirectly monitor the electrode degradation during welding by analyzing the electrode displacement signal from a non-contact sensor embedded in the welding machine and the electrode tip shape obtained from carbon imprint tests. As a result of an experimental campaign involving more than 1200 weld spots, the electrode speed during the final hold stage has been determined as the most explanatory feature describing the electrode displacement. Based on the mechanical strength of spot welds, the electrode contact face area has been defined as the most representative feature characterizing electrode degradation. A regression analysis has been carried out to infer a relationship between the electrode speed and the contact area representative of tool wear. A Neural Network has been built to use some features extracted from the electrode displacement signals to predict the contact area and thus indirectly the electrode degradation. The model has shown a good accuracy, with a mean error of the contact area about 1.61 mm2, and with a standard deviation of about 3.73 mm2. The data-driven approach proposed allows through the evaluation of the electrode contact area to have better real-time knowledge and control of the welding process.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call