After a long waiting period caused by the Corona pandemic, the conclusion of the modeling seminar could take place outside the university for the first time. In the heart of Franconian Switzerland, near Gößweinstein, the participants of the seminar presented their results to their supervisors and company representatives.
This project was a continuation from the 3rd modeling seminar of the Elite Graduate Program “Scientific Computing”. A produced spring was to be automatically compared with the previously produced CAD design. A 3D scan of the produced component served as the data basis.
The work in this project consisted of re-examining the algorithms developed in the previous year and adapting them better to the geometry of a spring. Ultimately, this resulted in a higher-quality overall result. By incorporating more efficient data structures, the comparison was also accelerated.
The focus is on the optimization of operating and component parameters in electronic circuits. Since the behavior of these circuits is very complicated, mathematical and programmatic methods are used for optimization. For these purposes, the circuits must first be mathematically modeled and then simulated.
The project involved optimizing the component parameters using a reinforcement learning algorithm. The optimization should additionally be improved by an extension of the used algorithm for different input and output variables.
The background to this subject is the fastening of a ship to the quay wall or the towing of a ship. In both cases, an elastic connection is used, for example in the form of an elastic rope. The aim of the project was the mathematical, time-dependent modeling, numerical simulation and visualization of an elastic rope using suitable numerical methods. Using a wide variety of case scenarios, it was possible to demonstrate how the developed model works.
