摘要
大规模模拟电路故障诊断,先用定位向量初定位故障子网络,再用神经网络定位故障子网络元件级故障。诊断分测前和测后,测前先将每个子电路的各种故障情况用Multisim模拟,求得各可及节点电压并作为BP的训练样本,再对所测电压信号分别归一化处理,定义网络的故障定位向量。最后用BP网络分析故障特征,进行故障诊断。
Large-scale simulation circuit fault diagnosis, first position the fault sub-network by dint of orientation vectors, then position the component fault of fault sub-network by dint of neural network. The diagnosis is divided into before test and after test, before test diversified fault of every sub-network should be simulated with the help of Multisim; then gain the voltage of tested node and take it as training sample of BP; once more unitary dispose the tested voltage signal and define the fault orientation vectors of network. Finally analysis the fault character by dint of BP network and diagnosis fault.
出处
《兵工自动化》
2007年第4期71-73,共3页
Ordnance Industry Automation
基金
总装司令部重点项目
关键词
模拟电路
故障诊断
定位向量
人工神经网络
故障子网络
元件级故障
Simulation circuit
Fault diagnosis
Orientation vectors
Artificial neural network
Fault sub-network
The component fault
