Project Details
Projekt
Physics-Informed Neural Operators for Generative Design of Tuned Vibration Absorbers in Urban Air Mobility
Applicant
Professor Dr.-Ing. Steffen Marburg
Subject Area
Mechanics
Term
since 2025
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 564325604
This research project aims to develop a data-driven design assistant for the generative design of tuned vibration absorbers in urban air mobility applications. Unlike traditional methods that rely on time-intensive iteration loops with manual parameter updating, the proposed design assistant can directly generate innovative design proposals based on predefined transmission spectra. At the core of this approach lies the method of neural operators, which learns the inverse mapping between a transmission spectrum and the corresponding design parameters. The project begins by developing analytical and numerical models of vibration absorbers, which form the foundation for generating the training data. Initially, purely data-driven neural operators are trained using these simulation datasets. To enhance prediction accuracy, the underlying physical equations are subsequently incorporated into the training process. Once trained, the neural operator is able to predict design parameters, such as the geometry of a vibration absorber, in real time based on a desired transmission spectrum. The workflow is tested on analytical and numerical simulation models before being validated with experimental data, including the quantification of prediction uncertainties. The resulting generative design assistant offers engineers a powerful, interactive tool for developing reliable design suggestions. It enables real-time analysis of the effects of design changes and allows for immediate adjustments, which significantly accelerates the entire design process. By combining engineering knowledge with the latest advances in artificial intelligence, the developed assistant facilitates robust and objective design decisions early in the development process. This marks a transformative step toward an innovative, data-driven design workflow, advancing the development of modern vibration absorbers for urban air mobility.
DFG Programme
Priority Programmes
