Description
Abstract: On the one hand, laser are more and more used in industries for material processing such as welding, cutting, drilling and even additive manufacturing. Such processes are highly dynamic, e.g. rapid heating, melting, solidification, cooling as well as keyhole processes. This makes the quality monitoring very challenging. On the other hand, acoustic emission (AE) systems are often used for process monitoring. Unfortunately, AE has two major drawbacks. First, the AE signals contain all kinds of information such as laser-material interaction, material transformation, defect creation and machine noise. Second, the correlation between the signals and the real events is often difficult to obtain. This is particularly true when an event takes place within the material as it is not directly visible. To overcome this difficulty, laser experiments were carried out at the European Synchrotron Radiation Facility (ESRF-Grenoble, France). Finally, we analyzed these AE signals by state-of-the-art machine learning technics to classify the laser process in terms of quality.
Authors: K. Wasmer
Keywords: Additive manufacturing, quality control, acoustic emission, high-speed X-ray imaging, machine learning