Dr. Jürgen LenzINNEO Solutions GmbH
Real-time data acquisition and forming process control using AI trained with simulation datasets
The process parameters and the respective quality information of components are recorded inline and in real time and transferred to a cloud-based data system. These quality information are determined by an optical measurement system (GOM Aramis) and laser triangulation (line scanner,gocator) and transferred to the ThingWorx enterprise IoT platform. For this purpose, an aging forming press, retrofitted with digital interfaces, is considered in order to support small and medium-sized companies with a toolchain to digitize the existing machinery. The captured data is correlated and evaluated using an Artificial Intelligence (AI) algorithm. In this process, the AI was trained using simulation data from a parameter study, experiments or a combination of both. In contrast to conventional training with exclusively genuine datasets, the results indicate that a comparable validation accuracy of the DNN-based quality assessment can also be achieved using simulation data generated by software scripts. This allows to significantly minimize both the testing effort as well as the production of real non-OK components and to increase the industrial applicability.