Transactions of China Electrotechnical Society  2024, Vol. 39 Issue (4): 1208-1220    DOI: 10.19595/j.cnki.1000-6753.tces.222148
Current Issue| Next Issue| Archive| Adv Search |
Evaluation Method for Power Transformer Winding Insulation Degradation Based on IoT Sensing Data and Tensor Fusion
Qu Yuehan, Zhao Hongshan, Cheng Jingyu, Ma Libo, Mi Zengqiang
School of Electrical and Electronic Engineering North China Electric Power University Baoding 071003 China

Download: PDF (1750 KB)   HTML (1 KB) 
Export: BibTeX | EndNote (RIS)      
Abstract  Online evaluation of winding insulation degradation is of great significance to the stable operation of transformers. Due to the lack of corresponding online sensing means, the traditional degradation evaluation methods relying on data such as furfural content and methanol content in oil cannot realize the online evaluation of winding insulation degradation. Most of the degradation evaluation methods based on online sensing data, such as voltage and current, only consider the influence of a single factor. It is difficult to fully reflect the degradation degree of winding insulation. Therefore, this paper proposes a power transformer winding insulation degradation evaluation method based on IoT sensing data and tensor fusion, considering the influence of electrical, thermal, and mechanical factors on insulation degradation. It relies on voltage, current, temperature, and partial discharge IoT sensing data to realize online evaluation of transformer winding insulation degradation.
Firstly, the cumulative damage mechanism of winding insulation degradation caused by electrical, thermal, and mechanical factors is analyzed. The transformer IoT sensing data of voltage, current, temperature, and partial discharge is used to construct the electrical, thermal, and mechanical performance degradation damage indicators of winding insulation. Then, based on tensor fusion, feature fusion of three degradation damage indicators is performed, and high-dimensional degradation feature correlation information between degradation damage indicators is extracted. Finally, the minimum quantization error of a self-organizing map is used to quantify the distance between the degradation feature output tensor and the best matching unit weight tensor. A comprehensive degradation evaluation index is then constructed, and the online evaluation of the winding insulation degradation degree is realized.
Based on the accelerated aging test data, the trend evaluation value, monotonicity evaluation value, robustness evaluation value, scale similarity evaluation value, and fusion evaluation value of the comprehensive degradation evaluation index are 95.78%, 100.00%, 99.75%, 93.24%, and 97.19%, respectively. The mean absolute, mean square, and root mean square errors of the comprehensive degradation evaluation index are less than 0.05. The R-Square and Pearson correlation coefficients exceed 97%, and the significance test coefficient is less than 0.01. Compared with traditional degradation evaluation methods, the fusion evaluation value of the proposed method exceeds 95% and has a higher similarity in trend with the furfural content index of winding insulation.
The conclusions are as follows: (1) The proposed method relies on the transformer IoT sensing data to quantify the cumulative damage caused by electrical, thermal, and mechanical stresses on winding insulation. It can accurately describe the degradation trend of winding insulation under multiple stresses. (2) The proposed method fuses the feature tensors of electrical, thermal, and mechanical degradation damage indicators based on tensor fusion, which can extract the high-dimensional degradation correlation information between degradation damage indicators to the greatest extent while retaining the characteristics of each degradation damage. (3) The proposed method can accurately describe the degree of deviation between the winding insulation degradation state and the healthy state through the comprehensive degradation evaluation index constructed by the minimum quantization error. (4) According to the experimental results, the proposed method can accurately evaluate the actual degradation state of the winding insulation based on the transformer IoT sensing data, and the evaluation results can provide a reference for the maintenance of the winding insulation.
Key wordsPower transformer      winding insulation      degradation evaluation      IoT sensing data      tensor fusion     
Received: 15 November 2022     
PACS: TM41  
Service
E-mail this article
Add to my bookshelf
Add to citation manager
E-mail Alert
RSS
Articles by authors
Qu Yuehan
Zhao Hongshan
Cheng Jingyu
Ma Libo
Mi Zengqiang
Cite this article:   
Qu Yuehan,Zhao Hongshan,Cheng Jingyu等. Evaluation Method for Power Transformer Winding Insulation Degradation Based on IoT Sensing Data and Tensor Fusion[J]. Transactions of China Electrotechnical Society, 2024, 39(4): 1208-1220.
URL:  
https://dgjsxb.ces-transaction.com/EN/10.19595/j.cnki.1000-6753.tces.222148     OR     https://dgjsxb.ces-transaction.com/EN/Y2024/V39/I4/1208
Copyright © Transactions of China Electrotechnical Society
Supported by: Beijing Magtech