面向换流变压器故障溯源的可解释知识图谱推理方法

吴婉豪; 王家文; 张永浩; 武宏斌; 李清泉; 任富强

doi:10.13296/j.1001-1609.hva.2026.09.018

您当前的位置：

首页 >

文章列表页 >

面向换流变压器故障溯源的可解释知识图谱推理方法

电力设备故障识别、预测与保护技术 | 更新时间：2026-09-10

- 面向换流变压器故障溯源的可解释知识图谱推理方法
- Interpretable Knowledge Graph Reasoning Method for Fault Traceability in Converter Transformer
- 高压电器 2026年62卷第9期页码：179-185,198
- 作者机构：
  
  山东大学特高压输变电技术与装备山东省重点实验室，济南 250061
- 作者简介：
  
  吴婉豪（2002—），女，硕士研究生，主要研究方向为变压器缺陷检测与诊断方法。
  任富强（1990—），男，博士，副研究员，主要研究方向为新型电力系统构建背景下电力设备多场耦合分析及可靠性提升、绝缘劣化机理、绝缘状态评估、人工智能在电力设备中的应用（通信作者）（E-mail：renfuqiang@sdu.edu.cn）。
- 基金信息：
  
  国家自然科学基金青年科学基金项目;Young Scientists Fund of the National Natural Science Foundation of China(52307180);山东省重大基础研究项目;Major Fundamental Research Project of Shandong Province(ZR2024ZD29)
- DOI：10.13296/j.1001-1609.hva.2026.09.018
  中图分类号：
- 收稿：2026-04-02，
  
  修回：2026-06-28，
  
  纸质出版：2026-09-16
- 稿件说明：
移动端阅览
吴婉豪, 王家文, 张永浩, 等. 面向换流变压器故障溯源的可解释知识图谱推理方法[J]. 高压电器, 2026,62(9):179-185,198.

WU Wanhao, WANG Jiawen, ZHANG Yonghao, et al. Interpretable Knowledge Graph Reasoning Method for Fault Traceability in Converter Transformer[J]. High Voltage Apparatus, 2026, 62(9): 179-185,198.
吴婉豪, 王家文, 张永浩, 等. 面向换流变压器故障溯源的可解释知识图谱推理方法[J]. 高压电器, 2026,62(9):179-185,198. DOI： 10.13296/j.1001-1609.hva.2026.09.018.

WU Wanhao, WANG Jiawen, ZHANG Yonghao, et al. Interpretable Knowledge Graph Reasoning Method for Fault Traceability in Converter Transformer[J]. High Voltage Apparatus, 2026, 62(9): 179-185,198. DOI： 10.13296/j.1001-1609.hva.2026.09.018.

摘要

换流变压器是高压直流输电系统的关键设备，其故障演化过程具有非线性和强耦合特征。针对运维文本数据利用不足和诊断过程缺乏可解释性的问题，提出一种可解释知识图谱推理方法（IKGR）。首先，采用ALBERT-BiLSTM-CRF从检修记录和故障报告等非结构化文本中抽取实体，并依据图谱模式层和物理因果约束组织关系三元组，构建换流变压器故障溯源知识图谱。其次，设计双智能体强化学习推理模型，引入软奖励机制，并利用短期同策略轨迹缓冲区同步更新双智能体参数，实现多跳故障溯源。实验结果表明， IKGR的根因排序指标

Hits

@10达到94.8%，并可生成包含设备部件、故障状态、保护动作和试验方法的推理路径，为换流变压器故障处置和运维决策提供路径级解释。

Abstract

Converter transformer is a critical component of high-voltage direct-current(HVDC)transmission systems

and its fault evolution exhibits nonlinear and strongly coupled characteristics. To address the problems of underutilized operation and maintenance text data and the lack of interpretability in the diagnostic process

an interpretable knowledge graph reasoning method(IKGR)is proposed. First

an ALBERT-BiLSTM-CRF model is adopted to extract entities from such unstructured text as maintenance records and fault reports. The relational triples are organized in accordance with the schema layer of the knowledge graph and physical causal constraints

thereby constructing a converter transformer fault traceability knowledge graph. Then

a dual-agent reinforcement learning reasoning model is designed

in which a soft reward mechanism is introduced and a short-term on-policy trajectory buffer is utilized to synchronously update the par

ameters of the dual agents

thereby achieving multi-hop fault traceability. Experimental results show that IKGR achieves 94.8% in terms of the root cause ranking meric

Hits

@10

and is capable of generating reasoning paths comprising equipment components

fault status

protective actions and testing methods

thereby providing path-level explanations for converter transformer fault handling and O

M decision-making.

关键词

Keywords

references

刘洪顺,隋宜臻,赵茹雪,等.换流变压器油纸绝缘气泡动态行为与高能电弧演化机理及其防治技术综述[J].高压电器,2026, 62(4):1-16. Liu Hongshun,Sui Yizhen,Zhao Ruxue,et al. Review on the evolution mechanism and active prevention of high-energy arcing induced by gas bubble dynamics in oil-paper insulation of converter transformers[J]. High Voltage Apparatus,2026,62(4):1-16.

陈洪波,曹培,高凯,等.换流变压器有载调压开关防爆试验研究与分析[J].高压电器,2024,60(2):119-126.

Chen Hongbo,Cao Pei,Gao Kai,et al. Research and analysis on explosion - proof test of converter transformerOLTC[J]. High Voltage Apparatus,2024,60(2):119-126.

汪佛池,程祥瑞,赵涛,等.油纸绝缘中气泡的生成特性及其对击穿性能的影响[J].高压电器,2020,56(1):61-67.

Wang Fochi,Cheng Xiangrui,Zhao Tao,et al. Characteristics of bubble formation in oil-paper insulation and its influence onthe breakdown[J]. High Voltage Apparatus,2020,56(1):61-67.

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.

Patil M,Paramane A,Das S,et al. Hybrid algorithm for dynamic fault prediction of HVDC converter transformer using DGA data[J]. IEEE Transactions on Dielectrics and Electrical Insulation,2024, 31(4):2128-2135.

Li H,Yao Q,Li X. Voiceprint fault diagnosis of converter transformer under load influence based on multi-strategy improved melfrequency spectrum coefficient and temporal convolutional network[J]. Sensors,2024,24(3):757.

孙长海,苏晓敏,李天伦,等.基于概率神经网络的油纸绝缘局部放电发展阶段识别[J].高压电器,2022,58(1):138-147.

Sun Changhai,Su Xiaomin,Li Tianlun,et al. Development stage recognition of partial discharge in oil-paper insulation based on probabilistic neural network[J]. High Voltage Apparatus,2022,58 (1):138-147.

汤健,侯慧娟,盛戈皞,等.变压器不平衡样本故障诊断的过采样和代价敏感算法[J].高压电器,2023,59(6):93-102.

Tang Jian,Hou Huijuan,Sheng Gehao,et al. Oversampling and cost-sensitive algorithm for transformer fault diagnosis with unbalanced samples[J]. High Voltage Apparatus,2023,59(6):93-102.

Tang Yachen,Liu Tingting,Liu Guang,et al. Enhancement of power equipment management using knowledge graph[EB/OL].[2026-04-02].https://arxiv.org/abs/1904.12242.

Tang Yachen,Han Haiyun,Yu Xianmao,et al. An intelligent question answering system based on power knowledge graph[C]//2021 IEEE Power & Energy Society General Meeting (PESGM),2021:1-5.

Li Cheng,Wang Bo. A knowledge graph method towards power system fault diagnosis and classification[J]. Electronics,2023,12 (23):4808.

Chaudhary A,Kafaie S,Rajabi E. EZ-check:explainable zero-shot knowledge extraction and fact checking using knowledge graphs[J]. Information Sciences,2026(742):123293.

Liu Hao,Li Dong,Zeng Bing,et al. Graph attention convolutional networks for interpretable multi-hop knowledge graph reasoning[J]. Information Processing&Management,2026,63(4):104581.

Zhang Xiaowen,Yan Li,Ma Zongmin. Multi-hop reasoning of knowledge chain over fuzzy spatiotemporal knowledge graphs with box embeddings[J]. Engineering Applications of Artificial Intelligence,2026(173):114443.

Wang Xiaowen,Han Huihui,Gao Benhe. Enhancing power transformer fault diagnosis through dynamic knowledge graph reasoning[J]. IEEE Transactions on Instrumentation and Measurement,2025 (74):1-9.

Ren Yuan,Nie Zhijie,Zhang Richong,et al. Progressive multi-hop reasoning for question answering over knowledge graphs[J]. Information Systems,2026(140):102721.

Lan Z,Chen M,Goodman S,et al. ALBERT:a lite BERT for selfsupervised learning of language representations[EB/OL].[2026-04-02].https://arxiv.org/abs/1909.11942.

Graves A,Schmidhuber J. Framewise phoneme classification with bidirectional LSTM and other neural network architectures[J]. Neural Networks:The Official Journal of the International Neural Network Society,2005,18(5/6):602-610.

Lafferty J D,McCallum A,Pereira F C N. Conditional random fields:probabilistic models for segmenting and labeling sequence data[C]//Proceedings of the Eighteenth International Conference on Machine Learning,2001:282-289.

Bordes A,Usunier N,Garcia-Durán A,et al. Translating embeddings for modeling multi-relational data[C]//Proceedings of the 27th International Conference on Neural Information Processing Systems,2013:2787-2795.

Lin X V,Socher R,Xiong C. Multi-hop knowledge graph reasoning with reward shaping[C]//Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing,2018:3243-3253.

Williams R J. Simple statistical gradient-following algorithms for connectionist reinforcement learning[J]. Machine Learning,1992,8 (3):229-256.

Xiong W,Hoang T,Wang W Y. DeepPath:a reinforcement learning method for knowledge graph reasoning[C]//Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing,2017:564-573.

Das R,Dhuliawala S,Zaheer M,et al. Go for a walk and arrive at the answer:reasoning over paths in knowledge bases using reinforcement learning[EB/OL].[2026-04-02].https://arxiv.org/abs/1711.05851.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

真空型有载分接开关谐波电流切换过程的电弧特性分析

换流变压器升高座电弧故障压力仿真分析

换流变压器油纸绝缘气泡动态行为与高能电弧演化机理及其防治技术综述

基于关系导向的电力设备故障缺陷文本实体及关系联合抽取方法

基于奇异值能量标准谱和改进TVF-EMD的换流变压器局部放电去噪方法