摘要
基于遗传编程(GP),提出了一种用于电力变压器绝缘故障诊断的判别函数法.该方法结合变压器油中溶解气体含量,利用GP算法的树状结构特点和模拟自然进化理论的全局寻优机制,自动从训练样本中学习到代表输入特征向量与对应故障类型之间关系的判别函数,以函数值的正负表示不同的故障类别.为了验证该方法的有效性,建立了变压器分层故障诊断模型,采用多个判别函数的方式逐步判别变压器绝缘故障的类型.与常规的三比值法、BP神经网络方法相比较表明,该方法提高了变压器绝缘故障诊断的正确率,具有良好的诊断效果.
A new kind of insulation fault diagnosis method for power transformers using genetic programming (GP) based discriminant functions was presented. Combined with the dissolved gas data, the method utilizes the structural feature of GP and its global search strategy to automatically construct a diseriminant function to stand for the relationship of input feature vector and corresponding fault types. The fault types are judged by the sign of function value. A hierarchical structure of transformer fault diagnosis model is built up to diagnose the fault type step by step by a set of discriminant functions. The proposed method was tested on the real gas records and compared with the conventional IEC method and artificial neural network method. The result shows that the proposed method has a good performance.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2006年第4期558-562,共5页
Journal of Shanghai Jiaotong University
基金
国家自然科学基金资助项目(50128706)
关键词
油中溶解气体分析
电力变压器
绝缘故障诊断
遗传编程
判别函数
dissolved gas analysis
power transformer
insulation fault diagnosis
genetic programming
diseriminant function