Name of the participant: Lucas Woltmann
Description of the IT research project: In production, especially automotive production, simulations are considered state-of-the-art for production optimization. Simulations help to make critical decisions in production planning and the production process. However, simulations are very time-consuming and, therefore, not suitable for time-critical problems. To compensate for this disadvantage, the core task for this project is to develop targeted AI-driven support for production processes as a data-driven simulation based on data properties, model quality, and domain expertise. In particular, this will be investigated for the complex systems in the field of automotive production. With the help of AI technologies, production processes will be improved, simplified, and accelerated. This project implements a hybrid approach of machine learning (ML) and simulation with respect to the model predictive control (MPC) process.
Within the KIKS project, individual components of a production simulation for queuing systems are replaced by ML models. The primary goal is to accelerate the existing simulation and not to replace the simulation with AI directly. The idea is to make a non-perfect prediction of the production throughput using ML models and then improve it with the simulation. In this way, we combine the simulation’s accuracy with the ML models’ efficiency to improve and accelerate production and production planning.
Software Campus partners: TU Dresden, Volkswagen AG
Implementation period: 01.01.2022 – 31.12.2023
© KIT.edu
