This project investigates how combining acoustic and vibration data can improve the detection and localization of faults in pumped-storage hydropower plants. These plants operate in dual modes—generation and pumping—which introduces dynamic stresses and accelerates wear on turbines. By designing a multimodal sensing pipeline and applying machine learning techniques, the goal is to identify the precise origin of anomalies and assess the added value of sensor fusion in real-world operational environments.
Thesis Type:
- Master Thesis / Bachelor Thesis
Goal and Tasks:
- Sensor Pipeline Design: Develop a synchronized data collection system combining acoustic and vibration sensors suitable for hydropower plant conditions.
- ML Model Benchmarking: Evaluate machine learning models for detecting and localizing anomalies using multimodal sensor data.
- Sensor Contribution Analysis: Assess the impact of vibration data on localization accuracy and identify strengths and limitations.
- • Extended Sensing Evaluation: Explore the potential of integrating additional sensor types (e.g., temperature, pressure) to enhance fault detection and localization.
Recommended Prior Knowledge:
- Signal processing and sensor integration
- Experience with machine learning and data analysis
- Programming skills (e.g., Python, MATLAB)
- Familiarity with industrial systems or energy technologies is a plus
- Interest in anomaly detection and predictive maintenance
Start:
- Flexible – ideally within the next 1–2 months / 6 months
Contact: