Artificial intelligence in machining
Artificial intelligence (AI) methods have already found their way into many areas of production research. This also applies to tool development, where AI-based force models support tool design, for example. These methods are based on empirical data, from which a model is formed using algorithms. Because these mathematical models are complex, they are often referred to as black boxes. In contrast, there is the deterministic model world, also called white box. If these model worlds are combined from data-driven approaches and physically describable contexts, a combined model world is created: the greybox. Using this method, which is to be researched, the qualification of coated tools for high-performance machining shall be made possible.
When predicting the tool performance, the currently applied methods lead to considerable deviations. On the one hand, this is due to the wear-related change in the shape of the tool that leads to progressive tool wear. On the other hand, the applied loads in the cutting process also change the coating and cutting material properties, which in turn influence the stress on the tool. These processes are very complex, so that they cannot be adequately described with simple models. "This is exactly where we want to use the potential of greybox models," explains Arnd Heckemeyer, head of the cutting department at IFW. "To represent the property changes of the cutting material with physical models would be a mammoth task. However, with the data-driven approaches, we have the appropriate tools to describe the effects and interactions of wear with tool load." While the overall focus of both projects is similar, the methods for data acquisition as well as the coupling with the deterministic whitebox model differ significantly. In one project, experimental data from planing studies are used to calculate the mechanical tool load and couple it to a blackbox model of time-resolved coating properties. The second project relies on a purely simulative approach in which tool wear is calculated taking into account the variable layer properties. Together with Institut für Werkstoffkunde (IW) from Hannover and Institut für Oberflächentechnik (IOT) from Aachen, they will work on the development of these model approaches in the first funding phase. "An exciting task that will be with us in the next few years and will lead to interesting findings," says Heckemeyer.
Contact:
For further information, please contact Arnd Heckemeyer, Institute of Manufacturing Engineering and Machine Tools at Leibniz Universität Hannover, by phone +49 (0) 511 - 5207 or by e-mail (heckemeyer@ifw.uni-hannover.de).
