Online Evaluation of Potential Transformer Operating Conditions Based on Similarity and Cluster Analysis

The paper proposes a comprehensive method for online monitoring of potential transformers (PTs) based on similarity coefficient clustering and health index evaluation. By collecting and analyzing real-time operational data, the method calculates similarity coefficients to assess the condition of PTs and employs cluster analysis to group transformers with similar characteristics. A health index model is developed to provide an accurate evaluation of PT status, enabling more effective maintenance and fault prediction. The approach has been verified through case studies, demonstrating its potential for early fault detection, improved PT management, and reduced maintenance costs in power systems.