模糊理论与神经网络结合对模拟电路进行分层故障诊断

The Diagnosis on Analog Circuits Based on Neural Network and Fuzzy Rules Neural Network

文中提出了用神经网络和基于模糊规则的神经网络联合对模拟电路进行分层故障诊断的方法。利用模糊规则将故障定位于某几个元件再用神经网络确定故障元件,针对某些数据对网络收敛不利的情况提出用神经网络进行数据的预处理。仿真结果说明这种神经网络模块化结构和分层测试方法不仅构造简单,便于进行模拟电路的自动测试,而且提高了模拟电路故障诊断的正确率。

In this paper, the neural network and the neural network based on fuzzy rules have been used together to diagnose a analog circuit. First the fault has been located in several resistances by fuzzy rules, then the exact fault resistace has been found by a small neural network. Neural networks have been used to amplify the small date. The enlarged data make the neural network easy to contract. The promising simulation result shows that they can accomplish the work even with simple structure and without complex math analysis. These methods can be easily used in automatic test.