A digital twin approach for vibration prediction of converter transformers

Read full text Summarise with VIRAT
Authors: Yao Yuan, Kejie Huang, Song Wang, Yichuan Ye, Kaixing Hong
Cited by: 0
References: 32

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

The converter transformer plays a crucial role in the DC transmission system. Unlike conventional transformers, its winding current is influenced by the switching actions of power converters, leading to the generation of high-order current harmonics. Predicting the vibrations of converter transformers presents significant challenges in equipment condition monitoring. This study introduces a novel digital twin approach based on the explainable neural network (XNN) model, which integrates the physical mechanisms of winding vibration responses with neural networks. The model updates its parameters using measured vibration signals, and establishes a nonlinear relationship between current harmonics and vibration harmonics, enabling precise vibration prediction in response to varying currents. In experimental validation, six identical converter transformers were selected. The XNN model was trained and verified using the vibration data collected during power decreasing and power increasing processes, respectively. After that, the digital twin approach was used to monitor the transformer vibration at arbitrary load. The results show that the proposed approach not only intuitively presents the physics-embedded nonlinear relationship between electromagnetic force and vibration, but also has higher accuracy in vibration prediction.

Fields of Study & Subjects

Power Transformer Diagnostics and Insulation Machine Fault Diagnosis Techniques Magnetic Properties and Applications

Publication details

Journal / source
Digital Twin
Publication date
26 March 2026
Language
EN