摘要
变压器油中溶解气体分析是电力变压器绝缘故障诊断的重要方法。文中将人工免疫网络分类算法应用于电力变压器故障诊断,利用增加抗原、记忆抗体类别信息的人工免疫网络对故障样本进行学习,可以获取更好地表征故障样本特征的记忆抗体集,再用最邻近分类法对故障样本进行分类。经大量实例分析,并将其结果与IEC三比值法和BP神经网络等方法的结果相比较,表明该算法能有效地对电力变压器单故障和多故障样本进行分类,具有较高的诊断准确率。
Dissolved gas-in oil analysis (DGA) plays an important role in fault diagnosis of power transformers. An artificial immune network classification algorithm is proposed for insulation fault diagnosis in this paper. To begin with, both antigens and memory antibodies with class information added to artificial immune network are trained to learn the feature of fault samples. In this way, memory antibody cells poll can represent the fault samples better than those obtained without class information. Then the k-nearest neighbor method is used to classify the fault samples. A mass of fault samples are analyzed in the algorithm proposed and the results are compared with those obtained by the IEC three-ratio method and BPNN. The comparison result indicates that the algorithm proposed has better classifying capability for single fault and multiple fault samples as well as high diagnosis precision.
出处
《电力系统自动化》
EI
CSCD
北大核心
2006年第6期57-60,共4页
Automation of Electric Power Systems
基金
国家自然科学基金资助项目(50425722)。~~
关键词
电力变压器
油中溶解气体分析
故障诊断
人工免疫网络
最邻近分类法
在线监测
power transformer
dissolved gas-in-oil analysis
fault diagnosis
artificial immune network
k-nearest neighbor method
online monitoring