摘要
首先简要介绍了基于油中溶解气体DGA(Dissolved Gas-in-oil Analysis)的变压器故障诊断机理,然后介绍了反向传播神经网络BPNN(Back-propagation Neural Network)的网络结构、学习算法和训练流程,并结合变压器故障实际特点,分析了输入输出模式的确定、隐含层设计、传递函数和训练函数的选择对于整个网络设计的重要性,通过在MATLAB中神经网络工具箱平台上的仿真比较找出合理的参数,从而建立基于BPNN的变压器故障诊断模型,最后通过对验证样本的仿真诊断结果对比,说明了该模型在实际应用中的有效性。
The paper presents fault diagnosis mechanism of a transformer based on dissolved gas-in-oil analysis, then presents the network structure of back-propagation neural network, learning algorithm and training processes. And combining actual characteristic of the transformer fault, analyze determination of the input and output mode, implication layer design and selection of training functions for the whole network design importance. By the simulation comparison of the tool-box platform of MATLAB neural network, find out reasonable parameters, building a transformer fault diagnosis model based on BPNN. Finally, by the simulation diagnosis result comparison of the verification sample, it is shown that the model is effective in practical use.
出处
《电气开关》
2012年第3期41-45,共5页
Electric Switchgear
关键词
油中溶解气体
反向传播神经网络
变压器
故障诊断
仿真参数研究
dissolved gas-in-oil analysis
back-propagation neural network
transformer
fault diagnosis
simulation parameters study