Deep Learning for Fault Detection, Classification, and Localization in Power Systems

Type: MA thesis

Status: open

Supervisors: Julian Oelhaf

Tasks:

  • Design and benchmark deep learning models (CNNs, RNNs, Transformers) for fault detection, classification, and localization in high-voltage power systems.

  • Work with high-resolution time-series data (current/voltage signals from simulations).

  • Investigate advanced concepts like knowledge distillation, transfer learning, and multi-task learning.

  • Analyze robustness to data scarcity, sensor dropout, and noise.

  • (Optional) Extend the pipeline for real-time or distributed inference.

  • (Optional) Co-author a scientific paper based on your results.

Requirements:

  • Strong programming skills in PyTorch

  • Experience with training deep learning models

  • Ability to attend in-person meetings

  • Bonus: Interest in signal processing, ML robustness, or time-series analysis

Application:

Send your application with the subject
“Application Fault DL Thesis + your full name” to julian.oelhaf@fau.de and include:

  • Curriculum Vitae (CV)

  • Short motivation letter (max. one page)

  • Transcript of records

 

This topic can also be conducted as a smaller project (e.g., research or programming project) instead of a full thesis.