Vibration Signal Analysis for On-Load Tap Changer Monitoring : A Signal Processing Framework for Tracking Mechanical Deviations

Sammanfattning

This thesis presents signal processing methods and an algorithm for monitoring the timing performance of On-Load Tap Changers (OLTCs) through vibration signal analysis. OLTCs are critical components in power transformers, and deviations in switching times can indicate early signs of incipient faults, unfavorable operating conditions, or mechanical wear. The proposed method detects characteristic vibration bursts during switching operations and extracts timing features to track long-term changes. Vibration data were collected from three different OLTC models using an accelerometer-based measurement system. Detailed analysis was carried out on OLTCs of the UCL and VUCG types. Reference voltage signals from the diverter switch were also recorded to reflect the mechanical switching sequence and serve as validation data. To detect anomalies, the method incorporates a scoring and flagging mechanism based on deviations from expected burst timings. The algorithm was evaluated under varying conditions, such as temperature shifts and mechanical wear. Results show a strong correspondence between the algorithm’s output and reference signals, supporting its potential for automated OLTC condition monitoring and predictive maintenance.