摘要
为提高电力设备状态评估的准确度,以变压器油为例,结合无监督学习算法设计了一种结合模糊C均值聚类算法的变压器油老化状态的评估方法。运用相对老化程度对单一指标进行量化,采用最优权重方法对单一指标加权求和得到组合指标的状态,进而得到油样本老化状态评估,对变压器油样本进行模糊C均值聚类分析,得到聚类中心和油老化程度的关系,验证模糊C均值分类可表示油样本老化分类,并在相关文献中找到实例验证油老化评估方法和模糊C均值聚类方法的有效性。
To improve accuracy in state evaluation of electric equipment, this paper takes the transformer oil for an example and designs a kind of evaluation method for transformer oil aging state based on the fuzzy C-means clustering method by combing the unsupervised learning algorithm. It uses relative aging degree to quantify the single index and adopts the optimal weight method for weighted sum and obtaining the state of combined indexes, and then gets aging state evaluation of oil samples. According to fuzzy C-means clustering analysis on transformer oil samples, it obtains relationship between the clus- tering center and oil aging degree and verifies the fuzzy C-means classification method could represent oil sample aging classi- fication. Meanwhile, it verifies effectiveness of the evaluation method for oil aging and the fuzzy C-means clustering method on the basis of actual examples in related literatures.
作者
曾振达
吴杰康
陶飞达
邹志强
罗伟明
张丽平
杨夏
ZENG Zhengda;WU Jiekang;TAO Feida;ZOU Zhiqiang;LUO Weiming;ZHANG Liping;YANG Xia(Heyuan Power Supply Bureau of Guangdong Power Grid Co.,Ltd.,Heyuan,Guangdong 525000,China;School of Automation,Guangdong University of Technology,Guangzhou,Gugndong 510006,China)
出处
《广东电力》
2018年第10期73-81,共9页
Guangdong Electric Power
基金
国家自然科学基金项目(51567002
50767001)
广东省公益研究与能力建设专项资金项目(2014A010106026)
广东电网有限责任公司科技项目(031600KK52160004)
关键词
变压器
油老化
状态特征评估
模糊C均值聚类
组合指标
transformer
oil aging
state characteristic evaluation
fuzzy C-means clustering
combined index
