Part-to-Part Control in Deep Drawing

Abstract

The aim of the thesis is to introduce and develop a part-to-part control algorithm which only uses the settings of the press as control variables. As a state variable, the draw-in is evaluated based on simulations. In addition, the simulations of a kitchen sink are chosen as a baseline model for the design of a possible control algorithm. For the baseline model, variant simulations are carried out in AutoForm and LS-Dyna to evaluate different aspects of the problem. In AutoForm, the influence of different noise parameters, such as friction or material changes, is evaluated, and the LS-Dyna model is used to increase the modelling accuracy of a blank holder system. Based on these variant simulations, different replacement models for the design of the control algorithm are created. Based on the replacement models, three different control algorithms are designed and tested in a virtual environment. The tested control algorithms range from simple proportional algorithms to purely model-based algorithms. Finally, the algorithms are partially tested and validated on two additional parts in the production line.