摘要
油中溶解气体分析(DGA)是判别变压器内部绝缘状况及发现内部潜伏性故障的重要手段。文中介绍了一种基于线性分类器、以DGA数据为特征参数的充油变压器潜伏性故障的识别方法。运用该方法进行了大量的应用实例分析,并将识别结果与BP神经网络法以及IEC三比值法进行了对比。结果表明选用H_2、CH_4、C_2H_2、C_2H_4、C_2H_6、CO、CO_2七种特征气体作为特征参数时,该方法显示出较高的准确度。
Dissolved gas analysis (DGA) is an important method to judge the insulation condition and find out the potential fault inside the oil-immersed transformer. A potential fault diagnosis method for oil-immersed transformer which is based on linear classifier (LC) and use the DGA data as the characteristic parameters is proposed in this paper. A large quantity of fault samples is analyzed by this method, and the results are compared with those obtained by BPNN and IEC three-ratio method.The pattern recognition results indicate that this method shows a higher accuracy, when the seven gases H_2、 CH_4、 C_2H_2、 C_2H_4、 C_2H_6、 CO、 CO_2、are chosen as characteristic parameters.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2004年第6期147-151,共5页
Proceedings of the CSEE
