AI-based setup assistance system for transfer presses

The Institute of Forming Technology and Forming Machines and the Institute of Measurement and Automatic Control are involved in the DFG Program SPP 2422 with a joint research project "AI-based setup assistance system for transfer presses".

Due to complex internal interactions with not fully identified relationships, the establishing and maintaining of stable process conditions in production on multi-stage presses can only be realized with a considerable adjustment effort and the utilization of implicit knowledge. If changes in the process parameters occur in one stage, which for example lead to changed process forces, this influences the process flow in other stages, which makes the reestablishment of good part production a laborious process depending on the complexity of the die set.

Within the scope of this project, a deeper understanding of the interrelationships of process influencing variables of multi-stage presses as well as the identification of significant influencing quantities with regard to the part quality is thus aimed at, by applying AI-based methods. For this purpose, a demonstrator component and suitable geometric quality variables are defined and a multi-stage forming process is designed. On the basis of a data acquisition system, the die set will be supplemented by various sensors in order to acquire the system variables occurring during the process (such as process forces, temperatures, structure-borne noise). Furthermore, the quality of the process is quantified by comparing the reconstructed and the optimal quality variables.

For the modeling of the process interdependencies, an approach consisting of two interacting AI models is developed, which on the one hand identify the system-inherent dependencies and on the other hand predict the geometric quality characteristics, depending on the systems configuration. On the basis of nonlinear optimization methods, which run through the two-stage AI model, a recommended course of action for an optimal machine setup, for the recovery of good part production, is derived.

                                                                                                                                         Published by Dietmar Friesen