基于邻域粗糙集与深度信念网络的油浸式变压器故障诊断  

Oil-Immersed Transformer Fault Diagnosis Via Combining Neighbor hood Rough Sets and Deep Belief Networks

作　　者：李水天 黄雪莜 田伟 蒋晶 李钧涛[3] LI Shuitian;HUANG Xuexiao;TIAN Wei;JIANG jing;Li Juntao(Guangzhou Power Supply Bureau of Guangdong Power Grid Co.,Ltd.,Guangzhou 510000,China;Xinxiang Aviation Industry(Group)Co.,Ltd.,Xinxiang 453003,China;College of Mathematics and Information Science,Henan Normal University,Xinxiang 453007,China)

机构地区：[1]广东电网有限责任公司广州供电局,广东广州510000 [2]新乡航空工业(集团)有限公司,河南新乡453003 [3]河南师范大学数学与信息科学学院,河南新乡453007

出　　处：《河南工学院学报》2024年第1期7-13,共7页Journal of Henan Institute of Technology

基　　金：河南省科技攻关计划(212102210140,172102210047)。

摘　　要：针对油浸式变压器的故障类型分类问题,提出一种基于邻域粗糙集与DBN的故障诊断模型。通过对变压器中故障气体进行无编码比值处理,得到了35种故障特征气体比值。利用相关性分析与领域粗糙集算法对所得气体比值进行特征选择,删去冗余以及对故障类型没有贡献的特征气体比值。将所得的9种特征气体比值作为输入变量,构建DBN诊断模型,实现了油浸式变压器的故障诊断。DGA数据上的实验结果验证了所提方法的有效性和准确性。In order to comprehensively explore fault information in transformer fault data and address the problem of fault type classification in oil-immersed transformers,a fault diagnosis model based on Neighborhood Rough Sets and Deep Belief Network(DBN)is proposed.Through non-coding ratio processing applied to fault gases within transformers,a total of 35 fault feature gas ratios are derived.These ratios undergo feature selection utilizing correlation analysis and the neighborhood rough set algorithm,aim at removing redundancies and irrelevant gas ratios that lack signi ficance in identifying fault types.The resultant 9 selected gas ratios are utilized as input variables for the construction of a DBN diagnostic model intended for fault diagnosis in oil-immersed transformers.The effectiveness and accuracy of the proposed method were confirmed through experimental validation using dissolved gas analysis(DGA)data.

关 键 词：油浸式变压器 邻域粗糙集 深度信念网络 故障诊断 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]