摘要
基于直流测试只能测试电路的静态工作状态,对敏感于信号频率的元件不能作完全测试这一情况,提出了交流测试下模拟电路硬故障状态特征的提取方法,使得样本具有非常好的典型性,便于网络诊断速度的加快和准确性的提高,以及通过在典型训练样本的基础上对每一状态增加两个容差范围内随机样点作为附加训练样本点,改善了诊断系统对容差下故障的诊断能力;实验证明,该方法是有实用价值的。
The acquiring method of the circuit characteristics is given that makes the samples＂ classic better, the speed of the diagnosis faster and the accuracy better. And the method that by adding random samples is provided based on the classic samples, to improve the fault diagnosis performance of the neural network under tolerative difference of the elements. A example was given, and prove it can increase the efficiency of the diagnosis.
出处
《计算机测量与控制》
CSCD
2006年第5期586-588,共3页
Computer Measurement &Control
关键词
电路
故障诊断
神经网络
交流测试
electrocircuit
fault diagnosis
neural network
AC test
