For the production of e-mobility components such as cable harnesses, battery cells, power electronics, etc., ultrasonic metal welding is well-established process of choice. These electrical applications require high quality for every single connection; single points of failure and no possibility of repair after installation or commissioning are state of the art. At present, the prevailing binding mechanisms and their sensitivity to the numerous process influencing variables like base material hardness, surface, and cleanliness are still the subject of research. In order to ensure sufficient quality despite the lack of process understanding, random destructive testing is carried out during ongoing production. The welding systems’ internal monitoring methods are currently not sufficient to make a prediction of the joint quality achieved. To determine process phases and extract features regarding the joint formation, the observation of process vibrations at the horn, anvil, and the components using laser-doppler-vibrometry, laser triangulation sensors or other suitable external measurement technology is common. These methods require external accessibility to the measurement position, not given in the industrial production environment. In this study, measurements of the high-frequency power signal of the welding system are conducted, and several machine learning models for quality prediction are set up. To ensure the robustness, several disturbances, e.g., changing material hardness and cleanliness, are taken into account. Thus, it will be evaluated to what extent an industrially suitable quality monitoring can be implemented by means of electrical measuring technology and how much more accurate such an external measuring system is compared to the possibilities already available in the welding system.

The friction stir welding (FSW) process is known as a solid-state welding process, comparatively stable against external influences. Therefore, the process is commonly used with fixed welding parameters, utilizing axial force control or position control strategies. External and internal process disturbances introduced by workpiece, gap tolerance, tool wear, or machine/tool inadequacies are rarely monitored, and conclusions about the weld seam quality, based on the recorded process data, are not drawn. This paper describes an advancement, improving on research into the correlation of process force feedback events or gradual force changes and the resulting weld seam characteristics. Analyzing the correlation between examined weld sections and high-resolution rate force data, a quality monitoring system based on an analytic algorithm is described. The monitoring system is able to accurately distinguish sound welds from such with internal (void) and external (flash) defects.

The production of electrical components for vehicles, battery systems, and power electronics components using ultrasonic metal welding is now well established. The industrial applications take advantage of the large joining surface and the associated excellent electrical and thermal conduction properties. However, this also prevents the weld quality from being monitored based on resistance measurements. For this reason, mechanical properties are used as a substituting quality criterion for process optimization and control. The type of mechanical testing depends on the component geometry, resulting in different quality features not necessarily comparable to each other. The ultrasonic metal welding process takes place in successive phases, which are characterized by the varying relative movements of the parts to be joined and the welding tools, prone to changes of material hardness, surface cleanliness, and geometric deviations. In this study, we investigate the influence of different mechanical quality features as target parameters for quality prediction on prediction accuracy and robustness. For this purpose, we consider the joint formation and the mechanical strength over the welding time for different quality parameters, fracture behaviors, and joint geometries. Supervised machine-learning quality prediction models are developed for the respective processes. The predictions are based on external and internal welding machine sensors. In addition to the sensor configuration, the importance of the individual process phases as a data source for the prediction accuracy is also investigated. We show that the overall process predictability strongly depends on the quality of the quality feature itself. Combining different sensor signals improves prediction quality. Furthermore, an analysis of the first process phases is sufficient for a robust quality prediction.

Welding aluminum castings is known for a high sensitivity for hydrogen-induced weld porosity. Research has shown that the porosity detected in mixed material resistance spot welds between die-cast and wrought aluminum is predominantly classified as solidification porosity. But, nugget penetration depths into the wrought aluminum alloy are low. In this work, the effect of a forging electrode force during welding current time t _i or after the welding current is shut off on weld spot characteristics is investigated, when resistance spot welding (RSW) mixed material joints between EN AC-43500-T6/T7 and EN AW-6082-T6. A current cut-off test based on a welding profile according to VDA 238–401 reveals a welding current time t _i between 60 ms ≤ t _i ≤ 80 ms as relevant for porosity detection in cross-section analysis. Based on this, different welding profiles are tested, and resulting weld spot characteristics are analyzed when applying chisel tests, cross-section analysis, quasi-static shear tensile tests, and scanning electron microscopy. The application of a forging electrode force reduces weld spot porosity in RSW significantly, while a current downslope leads to an increase. Further, the application of a forging electrode force during welding current time leads to a significant decrease in the standard deviation of s = 1.07 kN to s = 0.08 kN in shear tension force, although the mean shear tension force is increased marginally from 7.07 kN to 7.15 kN. Yet, an increase in weld penetration depth ≥ 20% into EN AW-6082-T6 is only achieved when reducing the initial electrode force to 5 kN and initiating forging electrode force during t _i .

Friction stir welding is a solid-state joining process that operates below the material’s melting point commonly used to join aluminum parts, avoiding the drawbacks of fusion-based methods. These resulting advantages have accelerated growth and are increasing the number of applications across a range of industrial sectors, many of which are safety–critical. Along with the increase in applications and rise in productivity the need for reliable and cost-effective, non-destructive inline quality monitoring is rapidly growing. This publication is based on the research group’s ongoing efforts to develop a capable generalized inline-monitoring solution. To detect and classify FSW defects, convolutional neural networks (CNNs) based on the DenseNet architecture are used to evaluate recorded process data. The CNNs are modified to include weld and workpiece-specific metadata in the classification. These networks are then trained to classify transient weld data over a wide range of welding parameters, three different Al alloys, and two sheet thicknesses. The hyperparameters are incrementally tuned to increase weld defect detection. The defect detection threshold is tuned to prevent false negative classifications by adjusting the cost function to fit the needs of a force-based detection system. Classification accuracies > 99% are achieved with multiple neural network configurations. System validation is provided utilizing a newly recorded weld dataset from a different welding machine with previously used parameter/workpiece combinations as well as parameter combinations and alloys as well as sheet thicknesses outside the training parameter range. The generalization capabilities are demonstrated by the detection of > 99.9% of weld defects in the validation data.