Identification of winding deformation and inter-turn short circuit faults in dry-type transformers based on a combined vibration/ultrasound method

Authors

• Shenjiong YAO, Faculty of Electrical Engineering, Shanghai University of Electric Power, Shanghai 200090, China
• Jialu o CHAI, Faculty of Electrical Engineering, Shanghai University of Electric Power, Shanghai 200090, China
• Yingjing CHEN, Faculty of Electrical Engineering, Shanghai University of Electric Power, Shanghai 200090, China
• Yiyang ZHOU, Faculty of Electrical Engineering, Shanghai University of Electric Power, Shanghai 200090, China
• Lingxuan ZHANG, Faculty of Electrical Engineering, Shanghai University of Electric Power, Shanghai 200090, China
• Zhousheng ZHANG, Faculty of Electrical Engineering, Shanghai University of Electric Power, Shanghai 200090, China

Keywords

one-dimensional convolution Transformer model; deep learning; time-frequency feature; feature fusion; ultrasound signal; vibration signal

Abstract

The insulation structure of dry-type transformers differs from that of oil-immersed transformers, and their oil-free nature makes it impossible to diagnose and forewarn inter-turn insulation faults in real time via chromatographic analysis of gas in oil. To solve this problem, a method based on a combined vibration/ultrasound method is proposed to identify winding deformation and inter-turn short circuit faults in dry-type transformers. First, the ultrasonic and vibration signals generated during winding deformation and inter-turn short circuits in dry-type transformers are collected, and the time-frequency features of the ultrasonic and vibration signals are extracted, including kurtosis, variance, mean, centroid frequency, root mean square frequency, and frequency entropy. Then, combined with the 1DCNN-Transformer, the collection, feature fusion, and classification of the time-frequency features of these two types of signals are achieved to improve the accuracy of fault identification. Finally, the proposed method is verified through simulation analysis. The research results show that the accuracy of the proposed method in identifying winding deformation and inter-turn short circuit faults reaches 98.22%, which is higher than that of traditional identification methods relying solely on vibration signals.

DOI

10.19781/j.issn.1673-9140.2026.02.030

First Page

338

Last Page

348

Recommended Citation

YAO, Shenjiong; CHAI, Jialu o; CHEN, Yingjing; ZHOU, Yiyang; ZHANG, Lingxuan; and ZHANG, Zhousheng (2026) "Identification of winding deformation and inter-turn short circuit faults in dry-type transformers based on a combined vibration/ultrasound method," Journal of Electric Power Science and Technology: Vol. 41: Iss. 2, Article 30.
DOI: 10.19781/j.issn.1673-9140.2026.02.030
Available at: https://jepst.researchcommons.org/journal/vol41/iss2/30