GBP人工神经网络在变压器故障诊断中的应用

Application of GBP Artificial Neural Network to Fault Diagnosis of Power Transformers

人工神经网络在电力变压器故障诊断中得到应用．本文利用广义误差反传（ＧＢＰ）人工神经网络建立了相应的故障诊断神经网络模型，对变压器的故障诊断进行了新的探索．这种方法能克服普通反向传播算法所存在的容易陷入局部极小点，对初值要求较高等缺点．实例诊断结果证明，这是一种比较理想的方法．

Artificial Neural Network(ANN) has been used for fault diagnosis of power transformers.Based on Generalized Error Back Propagation(GBP) algorithm,a neural network model for transformer fault diagnosis is presented.This method can overcome the main weaknesses of the back propagation algorithm which are liable to fall into local minimal value easily and to advance strict demands of initial value.The examples show that this method is effective.