1999-2002: VW foundation (Warnecke, Michael Wiese, Doris Maus, Momper, Udo Schwarz, Andre Sitz)
We cope with the impact of the cutting tools geometry on the overall performance (chip length and type, surface rawness, etc.) of the high-speed cutting process. In order to do so, as first step the cutting process has to be characterized properly according to certain cutting parameters. If based on physical principles one could think of constructing a straightforward model consisting of coupled equations of motion for the whole system workpiece-cutting tool. An exact modeling of course has to be done on an atomic level, resulting in a high-dimensional, i.e., large system that is hardly treatable. Experience shows that although degrees of freedom are rather large, often some statistical considerations may reduce the problem to a low-dimensional one. Assuming this to be valid for the cutting process we try to find a model for such a low dimensional but sufficient set of describing variables - an approximate state of the system. In particular we look at what the system produces and try to extract the essential system variables and the form of the model, i.e., we use an inverse approach. In order to learn something about what is going on behind, the modeling requires a mathematical description in parametric form, i.e., especially no neural networks. The state has to be specified by data analysis tools, like spectral or blind model identification techniques. Finally, according to the obtained model geometry parameters of the tool shall be optimized with respect to increased performance of the cutting process.