Prediction of Transient Magnetic Field of Transformer Inrush Current Fusing Multi-surce Data and CNN-Transformer

GAO Xiaona; ZHOU Xin; DENG Zhiqiang; LIN Xiaozhu; YANG Han

doi:10.13296/j.1001-1609.hva.2026.09.019

您当前的位置：

首页 >

文章列表页 >

Prediction of Transient Magnetic Field of Transformer Inrush Current Fusing Multi-surce Data and CNN-Transformer

Technology for Fault Identification, Prediction and Protection of Power Equipment | 更新时间：2026-09-10

- Prediction of Transient Magnetic Field of Transformer Inrush Current Fusing Multi-surce Data and CNN-Transformer
- High Voltage Apparatus Vol. 62, Issue 9, Pages: 186-198(2026)
- 作者机构：
  
  贵州电网有限责任公司遵义供电局，贵州遵义 563000
  贵州大学电气工程学院，贵阳 550000
- 作者简介：
- 基金信息：
- DOI：10.13296/j.1001-1609.hva.2026.09.019
  CLC：
- Received：08 April 2026，
  
  Revised：2026-06-11，
  
  Published：16 September 2026
- 稿件说明：
移动端阅览
GAO Xiaona, ZHOU Xin, DENG Zhiqiang, et al. Prediction of Transient Magnetic Field of Transformer Inrush Current Fusing Multi-surce Data and CNN-Transformer[J]. High Voltage Apparatus, 2026, 62(9): 186-198.
DOI：

GAO Xiaona, ZHOU Xin, DENG Zhiqiang, et al. Prediction of Transient Magnetic Field of Transformer Inrush Current Fusing Multi-surce Data and CNN-Transformer[J]. High Voltage Apparatus, 2026, 62(9): 186-198. DOI： 10.13296/j.1001-1609.hva.2026.09.019.

摘要

为解决变压器空载合闸时全域瞬态磁场难以现场实测、传统有限元分析计算效率低下、单一模型预测精度不足的问题，通过有限元仿真构建覆盖7组合闸角工况的560组有效样本数据集，采用随机森林完成输入特征重要性筛选，借助主成分分析将10 418维磁场数据压缩至300维，搭建CNN-Transformer混合预测模型，通过一维卷积提取励磁涌流局部瞬态特征，结合多头注意力机制捕捉全局时序依赖，实现全域瞬态磁场端到端快速重建。实验结果表明，该模型平均绝对误差低至0.004 2 T，平均绝对百分比误差为0.42%，决定系数达0.987，单样本预测耗时4.5 ms，计算效率较传统有限元方法提升超10

倍，极端工况下全空间平均绝对误差为0.003 2 T，最大误差不超过0.01 T，全工况下决定系数均不低于0.982。该模型预测精度高、泛化能力强、计算效率高，可有效实现变压器励磁涌流瞬态磁场快速精准预测，为变压器状态评估与保护整定提供技术支撑。

Abstract

To address the challenges of the difficulty in field measurement of the global transient magnetic field during no-load closing operation of transformer

the low computational efficiency of traditional finite element analysis

and the insufficient prediction accuracy of single models

a dataset comprising 560 valid samples with coverage of seven closing-angle conditions is constructed through finite element simulations. The random forest algorithm is employed to perform feature importance screening

and principal component analysis is utilized to compress the 10418dimensional magnetic field data to 300 dimensions. A CNN-Transformer hybrid prediction model is then established

where one-dimensional convolution is used to extract local transient features of the inrush current and the multi-head attention mechanism is adopted to capture global temporal dependencies

thereby enabling end-to-end rapid reconstruction of the global transient magnetic field. Experimental results show that the average absolute error of the model is as low as 0.0042 T

the average absolute percentage error is 0.42%

and a coefficient of determination(

)of is 0.987. The prediction time per sample is 4

.5 ms

which is more than 10

times faster than traditional finite element methods. Under extreme conditions

the global average absolute error is 0.0032 T

with the maximum error not exceeding 0.01 T. The coefficient of determination under all operating conditions is not less than 0.982. The proposed model exhibits high prediction accuracy

strong generalization capability

and high computational effi ciency. It can effectively achieve fast and accurate prediction of the transient magnetic field during transformer inrush current

providing technical support for condition assessment and protection setting of transformer.

关键词

Keywords

references

董理科,白鹭,武娜,等.基于知识图谱的电力变压器故障预测方法研究[J].高压电器,2022,58(11):151-159.

Dong Like,Bai Lu,Wu Na,et al. Research on fault prediction method of power transformer based on knowledge map[J]. High Voltage Apparatus,2022,58(11):151-159.

邱晟璇,王松,刘子瑞,等.基于FRA法和CNN的变压器绕组故障诊断研究[J].大电机技术,2025(3):108-116.

Qiu Shengxuan,Wang Song,Liu Zirui,et al.Research on fault diagnosis of transformer windings based on FRA method and CNN[J]. Large Electric Machine and Hydraulic Turbine ,2025(3):108-116.

郭文,黄著,周鸣中,等.基于CatBoost算法的空载变压器励磁涌流预测[J].电气传动,2026,56(2):34-42. Guo Wen,Huang Zhu,Zhou Mingzhong,et al. No-load transformer excitation inrush current prediction based on CatBoost algorithm[J]. Electric Drive,2026,56(2):34-42.

莫仕勋,杨皓.一种考虑J-A磁滞的变压器电磁暂态模型[J].实验室研究与探索,2023,42(3):119-124.

Mo Shixun,Yang Hao. A transformer electromagnetic transient model considering J-A hysteresis[J]. Research and Exploration in Laboratory,2023,42(3):119-124.

范贺明,刘亚楠,胥莹.变压器空载合闸冲击绕组松动信号智能检测方法[J].信息技术,2025(11):112-116.

Fan Heming,Liu Yanan,Xu Ying. Intelligent detection method for loose signal of no-load closing impact winding of transformer[J]. Information Technology,2025(11):112-116.

胡文旺,邓超平,江伟,等.计及变压器负载的和应涌流产生机理与仿真分析[J].电网与清洁能源,2025,41(10):76-83.

Hu Wenwang,Deng Chaoping,Jiang Wei,et al. Principle and simulation analysis of the transformer sympathetic inrush considering transformer load[J]. Power System and Clean Energy, 2025,41(10):76-83.

孔煜斌.电力变压器剩磁的仿真拟合及验证[J].兵器材料科学与工程,2025,48(2):150-156.

Kong Yubin. Simulation fitting and verification of power transformer remanence[J]. Ordnance Material Science and Engineering,2025,48(2):150-156.

Sun Y,Zhang L,Zhang Z,et al. Overview of research status of DC bias and its suppression in power transformers[J]. Energies,2022, 15(23):8842.

谭正凯,刘海光,李铁华,等.基于选相合闸控制的变压器励磁涌流抑制方法研究与应用[J].石油和化工设备,2025,28(12):70-73.

Tan Zhengkai,Liu Haiguang,Li Tiehua,et al. Phase-selective closing strategy for transformer inrush current mitigation:implementation and analysis[J]. Petro-Chemical Equipment,2025,28 (12):70-73.

肖湃,张海龙,杜志叶,等.特高压换流变压器励磁涌流选相合闸抑制方法研究[J].高压电器,2023,59(5):146-153.

Xiao Pai,Zhang Hailong,Du Zhiye,et al. Research on phase selection closing suppression method of exciting inrush current of UHV converter transformer[J]. High Voltage Apparatus,2023,59 (5):146-153.

宋玉洁,李振兴.基于动态可调电阻的变压器励磁涌流抑制方法[J].中国电力,2025,58(12):147-154.

Song Yujie,Li Zhenxing. Transformer excitation inrush current suppression method based on dynamically adjustable resistors[J]. Electric Power,2025,58(12):147-154.

王子伟,疏佳铭,陆斌,等.基于GASF及双输入AlexNet-SVM的变压器励磁涌流检测[J].机械与电子,2025,43(8):30-39.

Wang Ziwei,Shu Jiaming,Lu Bin,et al. Transformer inrush current detection based on GASF and dual-input AlexNet-SVM[J]. Machinery & Electronics,2025,43(8):30-39.

栾云飞,张禄亮,季天瑶,等.基于直方图和余弦相似度的光伏电站主变压器励磁涌流识别[J].电测与仪表,2025,62(7):181-189.

Luan Yunfei,Zhang Luliang,Ji Tianyao,et al. Histogram and cosine similarity based magnetizing inrush current identification for main transformer in photovoltaic power station[J]. Electrical Measurement & Instrumentation,2025,62(7):181-189.

龙启,杨旭,薛淑鹏,等.基于圆坐标表征法的换流变压器励磁涌流辨识研究[J].电力系统保护与控制,2025,53(9):118-129.

Long Qi,Yang Xu,Xue Shupeng,et al. Research on identification of converter transformer inrush current based on circular coordinate representation method[J]. Power System Protection and Control,2025,53(9):118-129.

韩志.基于多元信息融合的变压器励磁涌流识别研究[J].微型电脑应用,2023,39(12):215-219.

Han Zhi. Research on transformer inrush current identification based on multivariate information fusion[J]. Microcomputer Applications,2023,39(12):215-219.

马云龙,李秀广,周秀,等.基于高压真空快速断路器的变压器励磁涌流抑制技术研究[J].高压电器,2022,58(10):136-142.

Ma Yunlong,Li Xiuguang,Zhou Xiu,et al. Research on excitation inrush current suppression technology of transformer based on high voltage vacuum fast circuit breaker[J]. High Voltage Apparatus, 2022,58(10):136-142.

Peng Qingjun,Du Hantao,Zheng Zezhong,et al. Prediction of magnetic fields in single-phase transformers under excitation inrush based on machine learning[J]. Sensors,2025,25(13):4150.

Chu Wenchao,Wang Qiong,Yang Bo,et al. Analysis and study of excitation inrush current characteritics of high-impedance power transformer[C]//2025 2nd International Conference on Smart Grid and Artificial Intelligence (SGAI),2025:1592-1596.

Wang Jianting ,Gao Yonglin ,Liu Yaxing ,et al. Application of a multi-scale feature-enhanced transformer for segmentation in complex intercropping scenarios[J]. Smart Agricultural Technology, 2026(14):102137.

Qazi E U H,Almorjan A,Zia T. A one-dimensional convolutional neural network(1D-CNN)based deep learning system for network intrusion detection[J]. Applied Sciences,2022,12(16):7986.

Zhang Ping,Jia Yiqiao,Shang Youlin. Research and application of XGBoost in imbalanced data[J]. International Journal of Distributed Sensor Networks,2022,18(6):15501329221.

Ren Weiqing,Qu Yuben,Dong Chao,et al. A survey on collaborative DNN inference for edge intelligence[J]. Machine Intelligence Research,2023,20(3):370-395.

肖白,李森,焦明曦,等.基于霜冰优化组合模态分解及Informer的空间负荷预测[J].电力建设,2026,47(4):108-121. Xiao Bai,Li Sen,Jiao Mingxi,et al. Spatial load forecasting based on RIME-optimized combination modal decomposition and Informer[J]. Electric Power Construction,2026,47(4):108-121.

郑玉周,杜金龙,武永强,等.基于改进热路模型和PINN的变压器直流偏磁温升预测及绝缘寿命评估[J].电力电容器与无功补偿,2026,47(2):107-117. Zheng Yuzhou,Du Jinlong,Wu Yongqiang,et al. Prediction of DCbias temperature rise in transformer and insulation life assessment based on an improved thermal circuit model andPINN[J]. Power Capacitor &Reactive Power Compensation,2026,47(2):107-117.

张楚,王熠炜,张越,等.基于改进堆优化算法和TimesNet神经网络的锂电池剩余寿命预测方法[J].太阳能学报,2025,46 (11):98-108.

Zhang Chu,Wang Yiwei,Zhang Yue,et al. Lithium-ion battery remaining useful life prediction method based on improved heap optimization algorithm and TimesNet neural network[J]. Acta Energiae Solaris Sinica,2025,46(11):98-108.

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Fault Diagnosis of Transformer Based on FCM and Improved PCA

Numercial Simulation Study on the Effect of Different Transformer Pool Fire Areas on Valve Hall Fires in Converter Stations

Transformer Oil Temperature Prediction Based on Attention-gated Recurrent-temporal Covolutional Network

Variation Research on Dissolved Gas Generation of Overheating Faults in Oil-paper Insulation System

Fault Diagnosis Method of Transformer Core Looseness Based on ICEEMDAN-RCMDEand K-means

Related Author

QIN Yu

HUANG Huihong

FANG Jian

CHEN Yan

JIAO Yangyang

DU Yuge

XIANG Lin

LI Fei

Related Institution

Guangzhou Power Supply Bureau

Xuji Electric Co., Ltd.

State Key Laboratory of Coal Combustion, Huazhong University of Science and Technology

Department of Information Engineering, Gansu Iron and Steel Vocational Technical College

College of Electrical and Control Engineering, Xi'an University of Science and Technology

⁰