[1]王 抗,李 威,郭 涛,等.基于k-means的自适应换流变运行状态分层预警方法研究[J].机械与电子,2026,44(07):98-104.
　WANG Kang,LI Wei,GUO Tao,et al.Research on Adaptive Stratified Early Warning Method for Converter Transformer Operation State Based on k-means[J].Machinery & Electronics,2026,44(07):98-104.
点击复制

基于k-means的自适应换流变运行状态分层预警方法研究()
分享到：

《机械与电子》[ISSN:1001-2257/CN:52-1052/TH]

卷:
44
期数:
2026年07期
页码:
98-104
栏目:
电力控制
出版日期:
2026-07-25

文章信息/Info

Title:
Research on Adaptive Stratified Early Warning Method for Converter Transformer Operation State Based on k-means
文章编号:
1001-2257(2026)07-0098-07
作者:
王 抗; 李 威; 郭 涛; 张 祥; 杜晓舟
国网江苏省电力有限公司超高压分公司,江苏 南京 211106
Author(s):
WANG Kang; LI Wei; GUO Tao; ZHANG Xiang; DU Xiaozhou
(UHV Branch,State Grid Jiangsu Electric Power Co.,Ltd.,Nanjing 211106,China)
关键词:
换流变; k-means聚类; 分层预警; 风险指数; 多维监测数据
Keywords:
verter transformer; k means clustering; hierarchical early warning; risk index; multi dimensional monitoring data
分类号:
TM407
文献标志码:
A
摘要:
针对换流变压器运行工况复杂、监测指标多源异构及老化特征时变性强等问题,提出一种基于k-means的自适应换流变运行状态分层预警方法。首先构建包括油温、热点温度、振动加速度、局部放电幅值、DGA指标及绝缘含水率等在内的多维监测指标体系;随后提出组件子系统设备三级风险指数体系,实现热、电、机、老化多物理过程的层级化量化;在此基础上,引入无监督k means聚类构建自适应预警模型,通过簇中心偏离与工程知识映射获得“正常关注严重”三级预警结果,突破传统固定阈值方法难以适应设备差异与运行环境变化的局限性。案例结果表明,该方法能够有效区分不同退化模式,预警分层清晰,主成分分析降维后的类间分离度良好,可准确识别高风险簇,为换流变精益化运维及差异化检修策略制定提供技术支撑。
Abstract:
To address the challenges of complex operating conditions,multi-source heterogeneous monitoring metrics,and the significant time-variance of aging characteristics in converter transformers,this paper proposes an adaptive hierarchical early warning method for the operating state of converter transformers based on k-means clustering.Firstly,a multi-dimensional monitoring indicator system is constructed, encompassing parameters such as oil temperature,hot spot temperature,vibration acceleration, partial discharge amplitude,DGA indicators,and insulation moisture content.Subsequently,a three-level risk index system of component-subsystem-equipment is proposed to achieve the hierarchical quantification of multiple physical processes,such as thermal,electrical,mechanical,and aging.On this basis,an unsupervisedk-means clustering algorithm is introduced to construct an adaptive early warning model.The model obtains three level early warning results of “normal-concern-serious” through cluster center deviation and engineering knowledge mapping,breaking through the limitations of traditional fixed thresholdmethods that are difficult to adapt to equipment differences and changes in operating environments.Case study results show that this method can effectively distinguish different degradation modes,with clear early warning hierarchy and good inter class separation after PCA dimensionality reduction.It can accurately identify high-risk clusters,and provide technical support for the formulation of lean operation and maintenance and differentiated maintenance strategies for converter transformers.

参考文献/References:

[1] 李腾,樊培培,廖军,等.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 [J].高压电器,2025,61(11):221-230.

[2] Li Yuan,Zhang Qiaogen,Zhao Yi,et al.DC withstand test with partial discharge measurement of convertor transformers:problems and suggestions for improvement [J].IEEE Transactions on Dielectrics and Electrical Insulation,2017,24(2):1105-1109.
[3] 程炜超.基于PDC的换流变压器油纸绝缘多因子老化特性研究 [D].哈尔滨:哈尔滨理工大学,2021.
[4] Devadiga A A,Jayaram S H.Effects of repetitive fast transients on aging of transformer insulation [J].IEEE Transactions on Dielectrics and Electrical Insulation,2024,31(4):2189-2197.
[5] Liu Daosheng,Guo Zhengyang,Ding Jin,et al.Effect of thermal aging on electrical characteristics of modified pressboard insulation of converter transformer[C]∥ The 16th IET International Conference on AC and DCPower Transmission (ACDC 2020).New York:IEEE,2020:1708-1711.
[6] 冯鑫.换流变压器阀侧直流避雷器在线监测方法 [D].武汉:华中科技大学,2024.
[7] Su M S,Lin Y H,Yi C L,et al.Aging evaluation of transformer oil-paper insulation using partial discharge patterns and clustering methods [J].IEEE Access, 2025,13:181385-181395.
[8] 郑涛,何瑞,杨鑫慧,等.故障性涌流影响下换流变差动保护的可靠性风险评估 [J].电网技术,2021,45(11):4490-4497.
[9] 宗万里,刘香江,李嘉伟.换流变绕温油温传感器异常及可靠性分析 [J].中国设备工程,2025(3):7-10.
[10] 杨梅.浅谈在线监测系统在换流站的应用与维护[J].机电信息,2020(29):72-73.

备注/Memo

备注/Memo:
收稿日期:2025-12-30
基金项目:江苏电力超高压公司面向生产一线的科技项目(CGY-2025003)
作者简介:王 抗 (1986-),男,江苏南京人,博士,高级工程师,研究方向为特高压换流站设备安全稳定运行技术。
更新日期/Last Update: 2026-08-27