Robust estimation of clinch joint characteristics based on data-driven methods

Given a steadily increasing demand on multi-material lightweight designs, fast and cost-efficient production technologies, such as the mechanical joining process clinching, are becoming more and more relevant for series production. Since the application of such joining techniques often base on the ability to reach similar or even better joint loading capacities compared to established joining processes (e.g., spot welding), few contributions investigated the systematic improvement of clinch joint characteristics. In this regard, the use of data-driven methods in combination with optimization algorithms showed already high potentials for the analysis of individual joints and the definition of optimal tool configurations. However, the often missing consideration of uncertainties, such as varying material properties, and the related calculation of their impact on clinch joint properties can lead to poor estimation results and thus to a decreased reliability of the entire joint connection. This can cause major challenges, especially for the design and dimensioning of safety-relevant components, such as in car bodies. Motivated by this, the presented contribution introduces a novel method for the robust estimation of clinch joint characteristics including uncertainties of varying and versatile process chains in mechanical joining. Therefore, the utilization of Gaussian process regression models is demonstrated and evaluated regarding the ability to achieve sufficient prediction qualities.