🎙 About The Episode
Electric motors run hot - and knowing their internal temperature is mission-critical. But here’s the catch: you can’t place sensors where it matters most. So how do engineers protect magnets from demagnetization, maximize power, and extend motor life?
In this episode, Tobias Moroder, Data Scientist, and Georg Goeppert, Systems Engineer at Schaeffler, break down Thermal Neural Networks (TNNs) - a hybrid modeling approach that fuses physics with machine learning to accurately estimate temperatures inside E-Motors.
Here's what you’ll learn:
- Why virtual sensing is essential for modern electric drives
- How TNNs merge lumped-parameter thermal models with neural networks
- Why hybrid models outperform black-box AI in safety-critical systems
- How this method scales to batteries, power electronics, and full vehicle systems
- How Schaeffler’s TNN implementation works in MATLAB - now open-sourced
- A clear, practical deep dive at the intersection of AI, physics, and automotive engineering.
🔗 Explore MATLAB EXPO talk: https://de.mathworks.com/videos/therm...
🔗 Thermal neural network paper by Wilhelm Kirchgässner et al., Engineering Applications of Artificial Intelligence, vol 117: https://www.sciencedirect.com/science...
🔗 MATLAB implementation: https://github.com/wkirgsn/thermal-nn...
Connect with Tobias on LinkedIn: https://www.linkedin.com/in/tobias-moroder-327241261/
Connect with Georg on LinkedIn: https://www.linkedin.com/in/georg-g%C3%B6ppert-11a592206/
⭐️ Leave a Review
If you enjoy listening to the podcast, please do leave a 5-star review – ping me on my socials and tell me who you want to see next!
You can also Tweet @EngineeredM and tell me parts of the podcast you enjoyed the most. Feel free to share the podcast in your network
