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

ZHANG Jing'e; CHENG Zhongjie; LIU Bao

doi:10.13296/j.1001-1609.hva.2026.08.008

您当前的位置：

首页 >

文章列表页 >

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

Research & Analysis | 更新时间：2026-08-12

- Transformer Oil Temperature Prediction Based on Attention-gated Recurrent-temporal Covolutional Network
- High Voltage Apparatus Vol. 62, Issue 8, Pages: 58-66(2026)
- 作者机构：
  
  甘肃钢铁职业技术学院信息工程系，甘肃嘉峪关 735100
  西安科技大学电气与控制工程学院，西安 710054
- 作者简介：
- 基金信息：
- DOI：10.13296/j.1001-1609.hva.2026.08.008
  CLC：
- Received：30 April 2026，
  
  Revised：2026-06-18，
  
  Published：16 August 2026
- 稿件说明：
移动端阅览
ZHANG Jing'e, CHENG Zhongjie, LIU Bao. Transformer Oil Temperature Prediction Based on Attention-gated Recurrent-temporal Covolutional Network[J]. High Voltage Apparatus, 2026, 62(8): 58-66.
DOI：

ZHANG Jing'e, CHENG Zhongjie, LIU Bao. Transformer Oil Temperature Prediction Based on Attention-gated Recurrent-temporal Covolutional Network[J]. High Voltage Apparatus, 2026, 62(8): 58-66. DOI： 10.13296/j.1001-1609.hva.2026.08.008.

摘要

顶层油温是评价电力变压器热特性的重要指标，准确预测顶层油温变化趋势对评估变压器的负载能力和及时发现潜在的内部热故障具有重要作用。针对顶层油温预

测中精度低、特征选择过程繁琐和信息使用错误的问题，提出了一种基于时间卷积网络（TCN）、注意力机制（attention mechanism）和门控循环单元（GRU）的变压器顶层油温预测方法（Attention-GRU-TCN）。首先，在卷积中引入TCN，保证预测过程中的因果性，减少未来数据对预测过程的干扰。其次，注意力机制的引入，赋予关键数据特征更高的权重，取代复杂耗时的数据筛选过程，提高预测精度。最后，在某公司与传统的变压器顶层油温预测算法（例如，RNN、LSTM、GRU、BiLSTM、Transformer等）进行实验实测。结果表明，与RNN模型、LSTM模型、GRU模型、BiLSTM模型、Transformer模型相比，Attention-GRU-TCN模型的平均绝对误差（

MAE

）分别下降23.69%、15.88%、7.77%、29.35%、14.51%；均方误差（

MSE

）分别下降45.79%、31.34%、12.31%、43.32%、38.85%；决定系数（

）上升6.0%、3.10%、0.79%、12.30%、4.57%。说明该方法对变压器顶层油温的预测精度更高，有助于为潜在热故障预警提供参考并分析变压器负载能力。

Abstract

Top oil temperature is an important indicator for evaluating thermal characteristics of power transformer. Accurately predicting the variation trend of the top oil temperature plays a crucial role in assessing the load capacity of the transformer and promptly identifying potential thermal faults. In view of such issues as low accuracy

cumbersome feature selection process

and incorrect use of information in top oil temperature prediction

a transformer top oil temperature prediction method(Attention-GRU-TCN)is proposed based on temporal convolutional network (TCN)

attention mechanism

and gated recurrent unit(GRU). First

TCN is introduced into the convolution to ensure causality in the prediction process and reduce the interference of future data on the prediction process. Then

the introduction of the attention mechanism assigns higher weights to key data features

replacing the complex and time-consuming data filtering process

thereby improving prediction accuracy. Finally

the proposed method is experimentally validated at a company against traditional top oil temperature prediction models(e.g.

RNN

LSTM

GRU

BiLSTM

andTransformer

ect).The results show that compared with the RNN model

LSTM model

GRU model

BiLSTM model and Transformer model

the Attention-GRU-TCN model achieves a decrease in mean absolute error(

MAE

)by 23.69%

15.88%

7.77%

29.35%

and 14.51% respectively

a decrease in mean squared error(

MSE

)by 45.79%

31.34%

12.31%

43.32%

and 38.85% respectively;and an increase in the coefficient of determination(

)by 6.0%

3.10%

0.79%

12.30%

and 4.57% respectively. It is shown that the method has higher prediction accuracy for transformer top oil temperature

which is conducive to providing a reference for early warning of potential thermal faults and analyzing the load capacity of transformers.

关键词

Keywords

references

霍伟锋,李应光,陈斌,等.基于油浸式电力变压器顶层油温预测模型的多参数灵敏性分析[J].变压器,2021,58(11):16-20.

Huo Weifeng, Li Yingguang, Chen Bin, et al.Multi-parameter sensitivity analysis based on prediction model of top - oil temperature of oil-immersed power transformer[J]. Transformer, 2021,58(11):16-20.

张若愚,石峥,郭建,等. 基于数据质量提升的变压器顶层油温预测模型[J]. 山东电力技术,2025,52(12):113-122.

Zhang Ruoyu, Shi Zheng, Guo Jian, et al. Transformer top oil temperature prediction model based on data quality improvements[J]. Shandong Electric Power,2025,52(12):113-122.

王飞,蔡庆娜.基于AIoT云平台与CNN-LSTM的变压器油温预测研究[J]. 变压器,2025,62(2):21-26.

Wang Fei, Cai Qingna. Research on transformer oil temperature prediction based on AIoT cloud platform and CNN-LSTM[J]. Transformer,2025,62(2):21-26.

郭鹏鸿,唐治平,张光昊,等.基于双重注意力机制WOA-CNN-BiLSTM模型的变压器油温预测方法[J]. 变压器,2025,62(1):41-50.

Guo Penghong, Tang Zhiping, Zhang Guanghao, et al. Transformer oil temperature prediction based onWOA-CNN-BiLSTM model with dual attention mechanism[J]. Transformer,2025,62(1):41-50.

Liu Jingrui, Hou Zhiwen, Liu Bowei, et al. Mathematical and machine learning innovations for power systems: predicting transformer oil temperature with beluga whale optimization-based hybrid neural networks[J]. Mathematics,2025,13(11):1785-1819.

张帆,俞永军,蔡重凯,等.基于多时间尺度的干式空心电抗器匝间短路故障演化与寿命预测模型[J]. 高压电器,2025,61 (8):99-107.

Zhang Fan, Yu Yongjun, Cai Chongkai, et al. Fault evolution and life prediction model for inter-turn short circuit of dry air core reactor based on multiple time scales[J]. High Voltage Apparatus, 2025,61(8):99-107.

朱夏,陈颂,袁明瀚,等.基于CNN-LSTM风光荷预测的主动配电网双层扩展规划方法[J]. 高压电器,2025,61(5):218-227.

Zhu Xia, Chen Song, Yuan Minghan, et al. Double layer expansion planning method for active distribution network based on CNN-LSTM wind, solar and load prediction[J]. High Voltage Apparatus, 2025,61(5):218-227.

汤健,侯慧娟,王劭菁,等.基于策略梯度优化的变压器油中溶解气体预测模型[J]. 高压电器,2025,61(7):91-100.

Tang Jian, Hou Huijuan, Wang Shaojing, et al. Prediction model for dissolved gas in transformer oil based on policy gradient optimization[J]. High Voltage Apparatus,2025,61(7):91-100.

胡雪莹,李琳.基于全寿命周期成本和综合故障率的油浸式变压器经济寿命评估[J]. 智慧电力,2024,52(1):47-54.

Hu Xueying, Li Lin. Economic life evaluation of oil-immersed transformer based on life cycle cost and comprehensive failure rate[J]. Smart Power,2024,52(1):47-54.

林苑,赵晋斌,孙明琦,等.新型电力系统下基于物理信息LSTM网络的电力变压器状态评估方法[J]. 电力系统保护与控制,2025,53(14):133-141.

Lin Yuan, Zhao Jinbin, Sun Mingqi, et al. Power transformer condition assessment method based on physics-informed LSTM network in the context of new power systems[J]. Power System Protection and Control,2025,53(14):133-141.

张晓华,吕志瑞,孙云生,等.基于修正热路模型的油浸式变压器绕组热点温度计算研究[J]. 电测与仪表,2025,62(3):78-84.

Zhang Xiaohua, Lyu Zhirui, Sun Yunsheng, et al. Research of hot spot temperature calculation in oil-immersed transformer based on modified thermal model[J]. Electrical Measurement &Instrumentation,2025,62(3):78-84.

潘文霞,陈星池,赵坤,等.基于温度场计算的油浸式变压器热点温度仿真分析方法[J]. 电测与仪表,2025,62(3):71-77.

Pan Wenxia, Chen Xingchi, Zhao Kun, et al. Simulation analysis method of hot spot temperature of oil-immersed transformer based on temperature field calculation[J]. Electrical Measurement &Instrumentation,2025,62(3):71-77.

董雪情,荆澜涛,田瑞,等.基于LSTM模型的变压器顶层油温预测方法[J]. 电力学报,2023,38(1):38-45.

Dong Xueqing, Jing Lantao, Tian Rui, et al. Prediction method of transformer top oil temperature based on LSTM model[J]. Journal of Electric Power,2023,38(1):38-45.

赵军,高树国,何瑞东,等.变压器油中乙炔门控循环单元网络多步预测超参数优化方法[J]. 高压电器,2024,60(7):163-172.

Zhao Jun, Gao Shuguo, He Ruidong, et al. Hyper-parameters optimization method for multi-step prediction of acetylene in power transformer oil by gated cyclic unit network[J]. High Voltage Apparatus,2024,60(7):163-172.

谭志超,范竞敏,冯陆滔,等.基于模态分解和混合式CNN-GRUT的变压器油中溶解气体预测方法[J]. 电工电能新技术, 2024,43(7):80-90.

Tan Zhichao, Fan Jingmin, Feng Lutao, et al. Prediction method of dissolved gas in transformer oil based on modal decomposition and hybridCNN-GRUT[J]. Advanced Technology of Electrical Engineering and Energy,2024,43(7):80-90.

卫永鹏,苏益辉,王胜利,等.基于改进门控循环单元的变压器油中气体浓度预测[J]. 电气技术,2022,23(2):55-60.

Wei Yongpeng, Su Yihui, Wang Shengli, et al. Prediction of gas concentration in transformer oil based on improved gated recurrent unit[J]. Electrical Engineering,2022,23(2):55-60.

周春阳,李亚锦,邓光武,等.基于多维度分析的换流变预警技术研究与应用[J]. 电气工程学报,2021,16(4):151-158.

Zhou Chunyang, Li Yajin, Deng Guangwu, et al. Multi-dimensional analysis and early warning model of converter transformer[J]. Journal of Electrical Engineering,2021,16(4):151-158.

孟衡,张涛,王金,等.基于多尺度时空图卷积网络与Transformer融合的多节点短期电力负荷预测方法[J]. 电网技术,2024,48(10):4297-4305.

Meng Heng, Zhang Tao, Wang Jin, et al. Multi-node short-term power load forecasting method based on multi-scale spatiotemporal graph convolution network and transformer[J]. Power System Technology,2024,48(10):4297-4305.

樊清川,于飞,宣敏.基于优化Bi-LSTM模型的电力变压器故障诊断[J]. 计算机仿真,2022,39(11):136-140.

Fan Qingchuan, Yu Fei, Xuan Min. Power transformer fault diagnosis based on optimized Bi - LSTM model[J]. Computer Simulation,2022,39(11):136-140.

李腾,廖军,樊培培,等.基于自注意力机制与改进循环神经网络混合模型的特高压换流变压器顶层油温预测方法[J]. 现代电力,2024,41(6):1167-1175.

Li Teng, Liao Jun, Fan Peipei, et al. Forecasting method for ultra-high voltage converter transformer top-oil temperature based on selfattention mechanism and improved recurrent neural network hybrid model[J]. Modern Electric Power,2024,41(6):1167-1175.

高芬,苏依拉,牛向华,等.基于Transformer的蒙汉神经机器翻译研究[J]. 计算机应用与软件,2020,37(2):141-146.

Gao Fen, Su Yila, Niu Xianghua, et al. Mongolian-Chinese neural machine translation based on transformer[J]. Computer Applications and Software,2020,37(2):141-146.

周磊,竺筱晶.基于MA-CNN-LSTM和自注意力机制的单变量短期电力负荷预测[J]. 科学技术与工程,2024,24(22):9408-9416.

Zhou Lei, Zhu Xiaojing. Univariate short-term electrical load based onMA-CNN-LSTM-self attention[J]. Science Technology and Engineering,2024,24(22):9408-9416.

Hao Siyu, Lee D H, Zhao De. Sequence to sequence learning with attention mechanism for short-term passenger flow prediction in large-scale metro system[J]. Transportation Research Part C-Emerging Technologies,2019(107):287-300.

李梦阳,陈柳,史蒙,等.基于聚合模态分解和TCN-BiGRU的光伏功率预测模型[J]. 太阳能学报,2026,47(2):91-99. Li Mengyang, Chen Liu, Shi Meng, et al. Photovoltaic power forecasting model based on aggregate mode decomposition and TCN-BiGRU[J]. Acta Energiae Solaris Sinica,2026,47(2):91-99.

李泓钢,缪旻.基于ATT-LSTM网络的多核系统封装温度预测[J]. 电子器件,2025,48(5):1074-1081.

Li Honggang, Miao Min. Temperature prediction of multi-core system-in-package based onATT-LSTMnetwork[J]. Chinese Journal of Electron Devices,2025,48(5):1074-1081.

Baron N. Vanishing gradients in an artificial deep neural network: the impact of activation functions and optimization techniques[J]. Journal of Computer and Communications,2025,13(11):128-162.

陈聪,候磊,李乐乐,等.基于GRU改进RNN神经网络的飞机燃油流量预测[J]. 科学技术与工程,2021,21(27):11663-11673.

Chen Cong, Hou Lei, Li Lele, et al. Prediction of aircraft fuel flow based on recurrent neural network[J]. Science Technology and Engineering,2021,21(27):11663-11673.

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Day-ahead Photovoltaic Power Forecasting Model Based on Spatio-temporal Local Attention Network and LSTM

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

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

Transformer Fault Diagnosis Method Based on KELM Optimized by Improved Sparrow Search Algorithm

Related Author

SU Lianchao

FAN Jingyu

ZHI Detian

WU Kai

JIAO Yangyang

DU Yuge

XIANG Lin

LI Fei

Related Institution

State Grid Weifang Power Supply Company

State Grid Shandong Electric Power Company

Tianjin Guotou Jinneng Power Generation Co., Ltd.

State Key Laboratory of Alternate Electrical Power System With Renewable Energy Sources, North China Electric Power University

Xuji Electric Co., Ltd.

⁰