进化规划算法在变压器故障诊断中的应用

Application of Evolutionary Programming to Fault Diagnosis of Power Transformer

人工神经网络常用的 BP算法有易陷入局部极小、收敛速度慢等缺点 ,故利用进化规划来快速求取近年广泛应用于变压器故障诊断的人工神经网络的参数 ,完成该网络的训练 ,实现变压器故障诊断 。

ANN(Artificial Neural Networks) has been widely used in transformer fault diagnosis these years. Because of the drawbacks of BP algorithm such as local convergence and slow learning rate, this paper proposes the evolutionary programming algorithm to tune the network parameters, finish the training of ANN and realize the fault detection.