BachelorBearing fault detection on edge devices
Build a low-power pipeline (preprocessing + classical ML) that runs on an MCU and detects early-stage bearing faults from raw accelerometer data.
Requirements
- Background in Mechatronics / Electrical / Computer Engineering
- Comfortable with Python, basic signal processing
- Bonus: C/C++ on microcontrollers
Transfer learning across machines
Reduce the domain gap between simulated vibration data and real measurements using DANN, CORAL or MMD strategies.
Requirements
- Solid Python + PyTorch / TensorFlow
- Understanding of CNN/Transformer architectures
- Interest in domain adaptation literature
Digital twin for rotating machinery
Build FEA + multibody digital twins (ANSYS, MSC Adams) to generate labeled fault data for downstream AI training.
Requirements
- ANSYS / MSC Adams experience
- Strong mechanics fundamentals
- Comfortable bridging simulation and ML pipelines
