Proposed diagnostic methodology using the cross-correlation coefficient factor technique for power transformer fault identification

Abstract: This study investigates the impact of electrical parameter variation of a high-frequency transformer model on its
sweep frequency response analysis (SFRA) signature to help in classification and interpretation. The simulations have
been done using MATLAB and compared with the reference data. The results of SFRA measurements are repeatable
up to and beyond 1MHz. The proposed diagnostic methodology using the Cross-Correlation Coefficient Factor (CCF) is
used to identify the transformer faults. CCF used to measure the degree of relationship between two variables that
establish a relation between the predicted and actual data set. The results of this proposed methodology using the CCF
compared with existing Chinese Standard factor (CSF) indicate that, the proposed method is valid to identify the
transformer faults. Characteristics of the proposed scheme are fully analyzed by extensive MATLAB simulation studies
that clearly reveal that this method can accurately identify the transformer faults compared with CSF. And also does
not affected by different fault conditions such as transformer normal condition, Turn to Turn Fault for both HV, LV
sides, Axial Fault and/or Radial Faults on both sides, Short Circuit Fault between H.V and L.V Sides, Short Circuit to
Ground Fault for both HV, LV sides.

Related materials