Functional clustering methods for resistance spot welding process data in the automotive industry

In the automotive industry, quality assessment of resistance spot welding (RSW) joints metal sheets is typically based on costly and lengthy offline tests, which are unfeasible in full-scale production a large scale. However, massive industrial digitalization triggered by Industry 4.0 framework makes online measurements RSW process parameters available for every joint produced. Among these, so-called dynamic curve (DRC) recognized as full technological signature welds. Motivated this context, article intends to show potentiality practical applicability clustering methods data represented curves, and, general, functional data. way, task separating DRCs into homogeneous groups pertaining welds with common mechanical metallurgical properties can be performed without need arbitrary problem-specific feature extraction. We provide hands-on overview most promising methods, apply them collected during lab tests at Centro Ricerche Fiat. The identified emerge strictly linked wear status electrodes, which, turn, conjectured impact final quality. analysis code, developed software environment R , accompanied an essential tutorial ICOSAF project set containing DRC measurements, openly https://github.com/unina-sfere/funclustRSW/.