Cluster Analysis for Transformer On-load Tap Changer Contacts Fault Identification

In this paper, a cluster analysis-based contact fault identification method for transformer on-load-tap-changer (OLTC) is proposed, aiming at the accurate identification of different wear states of contacts by mechanical vibration signals and high-frequency discharge signals. The study adopts the K-means++ clustering algorithm to construct a contact fault diagnosis model by combining key parameters such as mechanical wear amount and contact resistance. The experimental results show that the model can effectively distinguish between mild, moderate and severe wear states and provide a reliable basis for fault identification. This study provides a new technical means for OLTC contact condition monitoring and fault diagnosis, which helps to improve the operational safety and reliability of the equipment.