提出了基于SOFM神经网络的容差模拟电路故障模糊诊断方法及其实现。该方法将网络撕裂法和SOFM神经网络相结合进行故障测试.并运用所设计的模糊逻辑神经网络系统判断测试条件,定位容差模拟电路的子网络级故障。仿真试验表明该方法故障定位精确度高。撕裂迅速,有利于大规模容差模拟电路故障诊断的实现。
On the basis of SOFM neural network, a new fuzzy diagnosis method for tolerance analog circuits fault is proposed. The method has integrated network decomposition and neural network, and one fuzzy logic neural network system is designed for judging test condition so as to realize the sub-network fault location in tolerance analog circuits. After emulation, the new method with SOFM neural network is testified to have high location accuracy and quick decomposition, also it can be used in large tolerance analog circuit fault diagnosis.