1411

To overcome the difficulty of the circuit fault diagnosis in analog circuits, which is hard to analysis and judgment by traditional method, a novel method based on frequency domain analysis and ELM neural network is proposed. Since the circuit frequency response curve can reflect the operating state and characteristics of the circuit, when a fault occurs, the frequency response characteristic curve will also change. By analyzing the changes in the transfer spectrum characteristics to determine whether it is in a fault state, this method has clear physical meaning and good robustness. The training process can be simplified by using ELM neural network. Through simulation and experiment, the results show that the method is sensitive and feasible for the circuit fault detection.