电力变压器故障诊断的人工神经网络方法

A Neural Notwork Approach to Power Transformer Fault Diagnosis

电力变压器故障诊断对变压器、电力系统的安全运行有着十分重要的意义．在介绍人工神经网络（ANN）和模糊理论的基本工作原理的基础上，针对变压器故障诊断的特点，运用油中溶解气体分析法（DGA），采用分块化的反向传播神经网络（BP），建立了变压器故障诊断的神经网络模型．通过训练和实际测试，表明了这一方法的有效性和可行性．

Diagnosis of power transformer abnormality is important for power system reliability.The basic principle of artificial neural network (ANN) and fuzzy set theory are described in this paper. Dissolved gas analysis are used according to the characteristics of transformer fault diagnosis. The neural network model based on transformer fault diagnosis is built by using moduler back propagation (BP). The results of training and usting show that the method is effective and available.