摘要
针对目前浅层机器学习理论在变压器故障诊断上精度不高以及大多数诊断方法参考的信息特征量单一的现状,提出一种基于AdaBoost-RBF算法与Dezert-Smarandache理论(DSmT)的变压器故障诊断方法。选择反映变压器故障信息的油中溶解气体、试验及产气率数据构成诊断参量空间,利用AdaBoost算法改进RBF神经网络算法,应用AdaBoost-RBF算法搭建并行的训练单元构造变压器故障诊断识别框架的基本信度赋值(BBA)。基于多源信息融合的思想,应用DSmT对基本信度赋值进行融合得到最终诊断结论,该理论克服了D-S证据理论无法融合高冲突性证据的局限性。对110kV变压器进行仿真实例分析,结果表明所提方法具有良好的实用性。
Aiming at the low accuracy of shallow machine learning theory in transformer fault diagnosis and the fact that most diagnostic methods only refer to a single information feature,a transformer fault diagnosis method based on AdaBoost-RBF algorithm and DSmT(Dezert-Smarandache Theory) is proposed. The dissolved gas in oil,test and gas production rate data,which can reflect the transformer fault information,are used to form the diagnostic parameter space. AdaBoost algorithm is applied to improve the RBF neural network algorithm. A parallel training unit is cons-tructed with AdaBoost-RBF algorithm to construct the BBA(Basic Belief Assignment) for the transformer fault recognition framework. Based on the idea of multi-source information fusion,the final diagnosis conclusion can be achieved by applying DSmT to fuse the BBA,which overcomes the limitations of D-S evidence theory that it is unable to solve the fusion problem of high conflict evidences. The case study of 110 kV transformer is carried out,the result shows that the proposed method has good practicability.
作者
刘云鹏
付浩川
许自强
李刚
高树国
董王英
LIU Yunpeng;FU Haochuan;XU Ziqiang;LI Gang;GAO Shuguo;DONG Wangying(Hebei Provincial Key Laboratory of Power Transmission Equipment Security Defense,North China Electric Power University,Baoding 071003 ,China;State Key Laboratoiy of Alternate Electrical Power System with Renewable Energy Sources,North China Electric Power University,Beijing 102206,China;School of Control and Computer Engineering,North China Electric Power University,Baoding 071003,China;State Grid Hebei Electric Power Research Institute,Shijiazhuang 050011 ,China)
出处
《电力自动化设备》
EI
CSCD
北大核心
2019年第6期166-172,共7页
Electric Power Automation Equipment
基金
国家自然科学基金资助项目(51407076)
中央高校基本科研业务费专项资金资助项目(2015ZD28,2018QN-076)
国家电网公司科技项目(5204DY170010)~~
