面向油色谱化学成分分析的变压器并行诊断研究  

Research on Parallel Diagnosis of Transformer Oriented to Analysis of Chemical Composition of Oil Chromatography

作　　者：郑博文 王秋雯 牛斌 陶宇 ZHENG Bo-wen;WANG Qiu-wen;NIU Bin;TAO Yu(Yuxi Power Supply Bureau,Yunnan Power Grid Co.,Ltd.,Yuxi Yunnan 653100,China)

机构地区：[1]云南电网有限责任公司玉溪供电局,云南玉溪653100

出　　处：《粘接》2020年第3期39-43,共5页Adhesion

摘　　要：针对变压器油色谱分析的智能化需求,提出一种基于大数据和三维场景的变压器油色谱诊断模型。为验证该模型的准确性,首先采用HDFS+HBase完成对诊断架构的搭建;然后引入Storm框架+OAO-RVM的并行化诊断框架;通过obj完成对挖掘结果的可视化。最后,搭建服务集群,对上述方案进行验证,结果表明本方案对变压器DGA数据的挖掘准确率高达95.37%,并可实现大规模的变压器油色谱数据的挖掘,表现出良好的性能。同时通过三维场景,可实时展示变压器的溶解气体成分与比值,为变压器故障的可视化诊断提供了参考。Aiming at the intelligent requirements of transformer oil chromatography analysis,a transformer oil chromatography diagnosis model based on big data and three-dimensional scene is proposed.In order to verify the accuracy of the model,HDFS+HBase is first used to complete the construction of the diagnostic architecture;then the Storm framework+OAO-RVM parallel diagnostic framework is introduced;and the mining results are visualized through obj.Finally,set up a service cluster and verify the above scheme.The results show that the scheme has a mining accuracy of 95.37%for transformer DGA data,and can achieve large-scale mining of transformer oil chromatography data,showing good performance.Finally,set up a service cluster and verify the above scheme.The results show that the scheme has a mining accuracy of 95.37%for transformer DGA data,and can achieve large-scale mining of transformer oil chromatography data,showing good performance.At the same time,through the three-dimensional scene,the dissolved gas composition and ratio of the transformer can be displayed in real time,which provides a reference for the visual diagnosis of the transformer fault.

关 键 词：溶解气体成分 Storm框架 三维场景 相关向量机 

分 类 号：TM41[电气工程—电器]