摘要
油中溶解气体阈值是判断变压器异常的重要指标,其判断阈值来自于历史数据分类统计,而选取分类条件是计算阈值的前提。针对高维数据阈值计算分类条件难以确定的问题,提出一种基于分类水平的油中溶解气体分类条件选取方法。该方法通过SS-FCM计算数据子类中心点,进而求取分类集几何中心及分类水平,根据分类水平判断分类条件优劣并进行分类,结合威布尔分布模型计算差异化阈值。研究发现分类水平高的条件统计威布尔函数拟合度较高,并且计算得到的阈值误报率较小,相比于标准阈值具有更高的准确性,有效识别变压器潜在过热缺陷。
The threshold of dissolved gas in oil,which comes from the historical data classification statistic,is an important rule to detect transformer abnormal status.Selecting classification method is the first step to compute thresholds.However,it is difficult to select the classification method for high dimensional data thresholds.On the basis of classification level,this paper proposed a classification method for dissolved gases in oil.Firstly,by adopting SS-FCM,the researchers calculated the data center and selected the classification method according to the classification level of sub class centers.Then differential thresholds were calculated by Weibull distribution model.With the selected classification method,the Weibull function has high fitting degree and low false alarm rate.Compared with the standard thresholds,the thresholds calculated by the selected classification method has higher sensitivity and can effectively identify the potential overheating fault of transformer.
作者
荣智海
齐波
张鹏
李成榕
陈玉峰
RONG Zhihai;QI Bo;ZHANG Peng;LI Chengrong;CHEN Yufeng(School of Electrical and Electronic Engineering,North China Electric Power University,Beijing 102206,China;State Grid Electric Power Research Institute of Shandong Power Supply Company,Jinan 250002,China)
出处
《华北电力大学学报（自然科学版）》
CAS
北大核心
2019年第1期17-24,共8页
Journal of North China Electric Power University：Natural Science Edition
基金
国家高技术研究发展计划(863计划)项目(2015AA050204)
