摘要
在利用集成神经网络进行模拟电路故障诊断时,经常会遇到训练样本庞大的情形。文中以进行多故障诊断为目的,提出了一种基于正交试验法和集成神经网络的模拟电路故障诊断方法。通过正交试验法,大大减少训练样本,同时保证了故障覆盖率。经算例证明,该方法可行、有效。
We often meet huge stylebook problems when using fault diagnosis for analog circuit by integrated neural networks. A method of fault diagnosis for analog circuit based on crossing experimentation method and neural network is put forward in this paper, it is used to diagnose multi-failure. It can reduce stylebook greatly through using crossing experimentation method,and ensure the fault rate. This method is innovative. The theoretical analysis and experimental results both show that this method is correct,ailable.
出处
《探测与控制学报》
CSCD
北大核心
2008年第B10期40-42,47,共4页
Journal of Detection & Control
基金
海装预研项目资助(编号不公开)
关键词
正交试验法
集成神经网络
模拟电路
故障诊断
crossing experimentation method
integrated neural networks
analog circuit
fault diagnosis
