电工技术学报  2019, Vol. 34 Issue (7): 1505-1515    DOI: 10.19595/j.cnki.1000-6753.tces.L80183
电力系统 |
考虑多元因素态势演变的配电变压器迁移学习故障诊断模型
杨志淳1, 沈煜1, 杨帆1, 蔡伟2, 梁来明3
1. 国网湖北省电力有限公司电力科学研究院 武汉 430077;
2. 国网电力科学研究院武汉南瑞有限责任公司 武汉 430074;
3. 国网新疆电力有限公司电力科学研究院 乌鲁木齐 830018
A Transfer Learning Fault Diagnosis Model of Distribution Transformer Considering Multi-Factor Situation Evolution
Yang Zhichun1, Shen Yu1, Yang Fan1, Cai Wei2, Liang Laiming3
1. Electric Power Research Institute State Grid Hubei Electric Power Co. Ltd Wuhan 430077 China;
2. State Grid Electric Power Research Institute Wuhan Nari Co. Ltd Wuhan 430074 China;
3. Electric Power Research Institute of Electric Power Company in Xinjiang Province Urumqi 830018 China
全文: PDF (1501 KB)  
输出: BibTeX | EndNote (RIS)      
摘要 针对配电变压器故障数据有限及数据过期的问题,提出一种考虑多元因素态势演变的配电变压器迁移学习故障诊断模型。首先,构建了配变运行状态评价指标体系,对指标状态量进行模糊二元量化,利用模糊Apriori算法挖掘其与故障之间的关联关系,提取诱导变压器故障的关键状态量。针对配变故障数据有限,引入Tanimoto系数,将有效的辅助故障数据迁移至目标配变,建立了基于信息迁移的配变故障诊断模型;针对配变故障数据过期,引入健康指数描述配变状态,将不同健康等级的辅助故障数据进行迁移,建立了针对数据过期的配变故障诊断模型。在此基础上,利用迁移学习算法TrAdaBoost对上述模型中目标与辅助故障数据的权重进行迭代求解,进而输出配变故障强诊断器。最后,根据配变故障数据进行算例分析,仿真结果表明,该文所建模型故障诊断精度高,具有比传统诊断器更强的泛化能力。
服务
把本文推荐给朋友
加入我的书架
加入引用管理器
E-mail Alert
RSS
作者相关文章
杨志淳
沈煜
杨帆
蔡伟
梁来明
关键词 故障诊断迁移学习配电变压器态势演变TrAdaBoost算法    
Abstract:Aiming at the problem of limited fault data and data expiration of distribution transformers, a transfer learning fault diagnosis model of distribution transformer considering multi-factor situation evolution is proposed in this paper. Firstly, an evaluation index system for distribution transformer status is constructed, and fuzzy binary quantification is performed on the state variables. The relationship between the state variables and the fault is explored by the fuzzy Apriori algorithm, and the key state variables that induces transformer fault is extracted. The Tanimoto coefficient is introduced for the limited fault data of distribution transformers, and the effective auxiliary fault data is migrated to the target distribution transformer, on this basis, the fault diagnosis model of distribution transformer based on information migration is established. The health index is introduced to describe the distribution status and the auxiliary fault data with different health levels is migrated because data has expired, on this basis, the fault diagnosis model of distribution transformers with expired data was established. the weights of the target and auxiliary fault data in the above model are iteratively solved by using TrAdaBoost, and then the fault diagnostic model is output. Finally, an example analysis is carried on the basis of the distribution transformer fault data, the simulation results show that the fault diagnosis accuracy of the model in this paper is high, and it has stronger generalization ability than traditional diagnosis model.
Key wordsFault diagnosis    transfer learning    distribution transformer    situation evolution    TrAdaBoost algorithm   
收稿日期: 2018-06-28      出版日期: 2019-04-17
PACS: TM407  
基金资助:国网湖北省电力有限公司重点科技攻关资助项目(52153217000T)
作者简介: 沈煜男,1983年生,硕士,高级工程师,研究方向为配电网运维管理及状态管控技术等。E-mail:totoshenyu@163.com
引用本文:   
杨志淳, 沈煜, 杨帆, 蔡伟, 梁来明. 考虑多元因素态势演变的配电变压器迁移学习故障诊断模型[J]. 电工技术学报, 2019, 34(7): 1505-1515. Yang Zhichun, Shen Yu, Yang Fan, Cai Wei, Liang Laiming. A Transfer Learning Fault Diagnosis Model of Distribution Transformer Considering Multi-Factor Situation Evolution. Transactions of China Electrotechnical Society, 2019, 34(7): 1505-1515.
链接本文:  
https://dgjsxb.ces-transaction.com/CN/10.19595/j.cnki.1000-6753.tces.L80183          https://dgjsxb.ces-transaction.com/CN/Y2019/V34/I7/1505