This project focuses on optimizing the reliability of industrial machinery through advanced sensor monitoring and automated calibration. During your thesis, you will investigate two main types of sensor failures: physical hardware anomalies and software-related drifts. By combining hands-on test bench measurements with data analysis, you will develop a system that distinguishes between these failures and automatically applies software-based corrections where possible. Your final concept will help develop self-calibrating sensor systems.
- During your thesis you will conduct physical measurements on a ball screw drive (BSD) test bench to record and analyze both healthy ("good") sensor data and hardware-related sensor failures ("bad"). You will analyze a provided, pre-detected dataset containing software-related sensor deviations (e.g., sensor drift).
- You will determine the criteria that distinguish a software-correctable drift from a physical hardware failure, making the sensor eligible for a purely software-based recalibration.
- Furthermore, you will research and evaluate various software-based recalibration techniques.
- You will design a robust concept for an automated workflow that, based on your classification, triggers the appropriate software-based recalibration and subsequently tests if the sensor is operating correctly again.
- Moreover, you will validate your concept using both your newly recorded hardware datasets and the provided software drift datasets.
- Finally, you will identify cases where software recalibration is not sufficient (specifically focusing on the recorded hardware failures) and propose a strategy for how the system should handle these physical anomalies (e.g., flagging for maintenance or triggering hardware replacement).