A kind of analog circuit was studied and modeled for fault diagnosis based on the neural networks

including the backward propagation(BP) neural network

the radial basis function(RBF) network and the support vector machine(SVM).We established three diagnosis models mentioned above

compared their diagnosis performances

and advanced a viewpoint that a different model corresponds to a different strategy during faults diagnosis.The simulating results show that

the SVM model has higher diagnosis precision and indicates a "high regarded as low" strategy with the RBF model

while the BP model presents as a "low regarded as high" strategy when dealing with the uncertain signals

which provides a basis for further theoretical study on the fault diagnosis of the analog circuits.