摘要
文章就人工神经网络ＢＰ模型进行故障诊断过程中输入样本情况影响诊断结果的准确率，提出了在原有神经网络输入节点的基础上再增加一特征输入节点，以反映输入样本数据大小的特征量的新方法。以电力变压器的故障诊断为例，证明了这种新方法具有更高的故障诊断准确率。
In the fault diagnosis process using BP network and algorithm the accurate rate of diagnosis result is affected by the condition of input samples. This paper proposed a new method in which an adequate node for BP network is added to original neural network to indicate the real value of input samples. Taking the diagnosis of power transformer for example, the result of test shows that this new method has more accurate rate of fault diagnosis.
出处
《电网技术》
EI
CSCD
北大核心
1998年第11期62-63,68,共3页
Power System Technology
关键词
人工神经网络
电力变压器
故障诊断
Artificial neural network Data processing Adequate input node Power transformer
