基于多频测试和神经网络的模拟电路故障诊断

多频测试使模拟电路响应的故障状态和正常状态差异最大化,而神经网络具有解决复杂分类问题的能力。结合两者优点,提出一种基于多频测试和神经网络的故障诊断方法:通过灵敏度分析指导多频测试矢量生成,选择最优测试激励;提取各测试节点响应的故障信息,利用神经网络对各种状态下的特征向量进行分类决策,实现对故障元件的检测和定位。实验结果表明,该方法对模拟电路故障诊断非常有效,具有很强的实用性。

Fault diagnosis of analog circuits based on multifrequency test and neural networks

Multifrequency test can maximize differences between the failure state and the normal state of the analog circuit's response,and Neural Networks(NNs)have the ability to solve complex classification problems.An efficient approach for diagnosing faults in analog circuits is presented.It is based on the advantages of both multifrequency test and NNs.The sensitivity analysis is used to generate and choose the multifrequency test vectors of the Circuit Under Tes(tCUT).Fault features of the test point in CUT are extracted and fused.NNs are used to classify the features in a variety of state for the detection and location of faulty components in CUT.The experimental results show that this method is very effective and highly practical for fault diagnosis of analog circuits.