Lehrstuhl für Fertigungstechnologie, Universität Erlangen-Nürnberg



Detection and Adaptive Correction of Weld Paramter Disturbances using the Time-Variant Transfer Dynamics of the Laser Beam Welding Process



Project Status: finished

Members


Sensor signals of laser material processing are in general ultidimensionally correlated with process variables non-observable in-process. The quantification of these process variables using sensor signals therefore requires decorrelation of superposed effects. To be generally independent from combinations of work piece material, the laser machine, sensor arrangement and even the laser materials process, the decorrelation method has to be self learning and free of empiricism. Hence, the basics of a self learning Multiple Input Multiple Output model were researched, which is capable to decorrelate relationships connected by linear differential equation systems in real time. With respect to interpolation, convergence and robustness against additive noise, the performance of the model was evaluated and optimized. Aktuators


Letztes Update: 01.01.2010