Soft Fault Diagnosis of Analog Circuits Based on a ResNet With Circuit Spectrum Map
Lipeng Ji, Chenqi Fu, Weiqing Sun
Abstract
Deep learning has achieved excellent results in many fields due to powerful feature extraction and learning ability. In this study, an improved method for analog circuit fault diagnosis based on a deep residual network is presented. The proposed method utilizes a ResNet to extract the performance characteristics of an analog circuit and determine the fault type of a component to realize the fault diagnosis of a circuit. The Short-time Fourier Transform is used to convert the time-domain output signals of a circuit into two-dimensional circuit spectrum maps, which are further used as the ResNet input. The fault diagnostic performance of the proposed method is verified by simulation with the Sallen-key bandpass filter circuit and the Four-opamp biquad high-pass filter circuit. The simulation results show that the proposed method performs well on both test circuits, achieving the diagnostic accuracy of up to 99.1% on the second-mentioned circuit.