摘要
为了提高变压器故障诊断的准确率,提出一种免疫RBF混合智能诊断算法,用免疫聚类算法确定RBF神经网络隐含层中心的数量和初始位置,减少了网络训练的计算量,提高了网络的泛化能力;用遗传算法对RBF网络训练,进一步优化网络的结构和连接权重,将训练后的RBF网络应用于变压器故障诊断。经过大量实例分析,并将其结果与其他算法进行对比,表明该方法算法精简,诊断正确率高。
In order to improve the correct rate of dissolved gas analysis（DGA）,this paper investigates an intelligent mixed method of the artificial immune RBF neural network.First it determines number and initial position of RBF neuron in hidden layer with immune clustering algorithm,which can reduce the training computational complexity of neural networks and improve capacity of network generalization.Then it trains the RBF network with genetic algorithm to furtherly optimize the structure and connection weights of the network and applies the RBF neural network to transformer fault diagnosis.E xamples analysis and comparison results show that the precision of fault diagnosis can be evidently improved.
出处
《电力系统保护与控制》
EI
CSCD
北大核心
2010年第11期6-9,14,共5页
Power System Protection and Control
关键词
变压器
故障诊断
免疫聚类
RBF网络
power transformer
fault diagnosis
artificial immune clustering
RBF neural network
