广东电力2017，Vol.30Issue(9)：104-110,7.DOI:10.3969/j.issn.1007-290X.2017.09.019
基于聚类多级决策SVM的油纸绝缘沿面放电阶段识别方法
Identification Method for Surface Discharge Stage of Oil Paper Insulation Based Cluster-multistage SVM
摘要
Abstract
Based on surface discharge model for oil paper insulation rod-plane,constant voltage method was adopted for ex-periment research on development characteristic of partial discharge (PD)of this model under AC voltage. Four graphs of phase-resolved partial discharge (PRPD)modes at different time and 29 statistic characteristic quantities were extracted, kernel principle component analysis method was used for extracting eight new characteristic quantities and discharge develop-ment stages were divided by system clustering analysis as well. On the basis of eight new characteristic parameters,class dis-tance point in clustering analysis was used to form a hierarchical identification method for discharge stages based on cluster-multistage support vector machine (SVM). Results indicate this method based on cluster-multistage SVM can identify dis-charge stages well and get result consistent with that of system clustering analysis method.关键词
油纸绝缘/局部放电/发展特性/支持向量机/阶段识别Key words
oil paper insulation/partial discharge/development characteristic/support vector machine/stage identification分类
信息技术与安全科学引用本文复制引用
杨贤,林春耀,马志钦,饶章权,郑晓光..基于聚类多级决策SVM的油纸绝缘沿面放电阶段识别方法[J].广东电力,2017,30(9):104-110,7.基金项目
国家自然科学基金创新群体基金项目(51321063) （51321063）
中国南方电网有限责任公司重大科技专项(KJ-2014-170-3) （KJ-2014-170-3）
