An Online Data-Driven Technique for the Detection of Transformer Winding Deformations

Tianqi Hong, Digvijay Deswal, Francisco De Leon

Research output: Contribution to journalArticlepeer-review


This paper presents a novel online diagnostics method capable of detecting winding deformations in two-winding single-phase transformers. The main idea is to identify changes in the short-circuit impedance. The combination of 3-D Lissajous curve methods with a Butterworth low-pass filter allows for the accurate determination of winding deformation of large power transformers in real time. The method is very robust and capable of detecting deformations at the early stage even when the measurements are noisy. Only information already available to the differential protection relay is needed. The proposed diagnostics method has been validated with circuit and finite-element simulations plus a lab experiment. The results show that the proposed online diagnostics technique has the ability to identify winding deformation problems under severe conditions, such as nonsinusoidal input, nonlinear loading, and measurement noise. Under ideal conditions (no signal noise), the inductive identification error of the proposed online diagnostics method identifies the parameters with less than 0.09% error. When accepting a measurement noise of 1%, the error on the identification of inductance is less than 0.13%.

Original languageEnglish (US)
Pages (from-to)600-609
Number of pages10
JournalIEEE Transactions on Power Delivery
Issue number2
StatePublished - Apr 2018


  • Lissajous curve methods
  • measurement noise
  • non-sinusoidal excitation
  • nonlinear loading
  • two-winding single-phase transformers
  • winding deformation detection

ASJC Scopus subject areas

  • Energy Engineering and Power Technology
  • Electrical and Electronic Engineering


Dive into the research topics of 'An Online Data-Driven Technique for the Detection of Transformer Winding Deformations'. Together they form a unique fingerprint.

Cite this