摘要
基于神经网络的模拟电路故障诊断方法,存在训练样本庞大的缺点,以进行多故障诊断为目的,提出一种基于正交试验法和神经网络的模拟电路故障诊断方法,充分利用各种信息,以科学的方法对数量庞大的训练样本进行"缩水"处理。该方法能大大减少训练样本,同时保证故障覆盖率,有一定创新性。经算例证明,该方法可行有效。
The method of fault diagnosis for analog circuit based on neural network exits a shortcoming for huge stylebook. For the multiple faults diagnosis, put forward a new fault diagnosis of analog circuit based on crossing experimentation and neural network, and then using scientific method process shrink manage to the huge stylebook get across various information is ample. The method can reduce stylebook, ensure the fault rate and it is innovative. Illustration shows the method is effective.
出处
《兵工自动化》
2009年第9期74-75,84,共3页
Ordnance Industry Automation
关键词
正交试验法
神经网络
模拟电路
故障诊断
Crossing experimentation method
Neural networks
Analog circuit
Fault diagnosis
