AI-augmented Simulation

Enhancement of Simulation Methods by Artificial Intelligence and Machine Learning

The research group AI-augmented Simulation combines a wide range of physical-based simulation approaches and software tools with both mathematically advanced and efficient algorithms and AI frameworks to model and optimize power electronic devices and systems.


  • Coupled electrical, thermal and mechanical simulations in the domain of micro- and power electronic devices and energy systems
  • Implementation of customized software packages to increase efficiency, performance and functionality of standard software products
  • Simulation of electrical components based on CAD data from 3D CT scans

Artificial Intelligence

  • Clustering unknown data sets using unsupervised learning algorithms
  • Recurrent neural networks for analyzing and predicting time series
  • Convolutional neural networks for signal and image processing


  • Topology optimization of inductive components using gradient-based methods such as SIMP
  • Optimization of electrical networks and systems with genetic algorithms
  • Sensitivity analysis and visualization of big data


Physics-informed Neural Networks (PINNs) for Engineering Applications


PINNs seamlessly integrate physical principles into the learning processes of neural networks, enabling accurate modeling of complex systems as well as inter- and extrapolation based on fundamental laws.

THE high-speed Litz Wire Calculator for power losses


LiWiCalc® is the fastest and most accurate software calculating all dominant power losses on realistic litz wire geometries:

Highly accurate resolution of internal and external proximity as well as skin losses based on the latest scientific method SEEC (Strand-Element-Equivalent-Circuit).


Optimization and AI in Power Electronics


Optimization of Complex Heat Sink Structures


Simulation Analysis Based on CT Scan Data


AI-augmented Simulation


of Litz Wire


Publications AI-augmented Simulation