A novel analog circuit fault diagnosis method based on multi-channel 1D-resnet and wavelet packet transform

Abstract

To quickly and accurately locate the fault location and fault parameter deviation of analog circuits, a novel incipient fault diagnosis method based on multi-channel one-dimensional residual networks (MC-1D-ResNet) and wavelet packet transform(WPT) is proposed in this paper. The WPT is employed to preprocess the time-domain response signals of analog circuit, and the proposed MC-1D-ResNet is utilized for feature mining and fault classification.The two-level WPT is first carried out on the time-domain response signal to generate one approximate signal and three detailed signals. Secondly, MC-1D-ResNet further performs feature mining on approximate signals and three detailed signals, and realizes fault diagnosis. Through simulation analysis, the proposed method is fully evaluated with the Sallen-Key bandpass filter circuit and the four-op-amp biquad high-pass filter circuit. Even in complex Four-op-amp biquad high-pass filtering circuits, the diagnostic accuracy can reach 99.74%. This article also designs a hardware testing platform based on FPGA, and conduct actual fault diagnosis tests on the four-op-amp biquad high-pass filter circuit. The results show that the average accuracy of 50 actual diagnoses for each type of fault in the circuit was 97.80%.