A novel approach for analog fault diagnosis based on LMD decomposition and reconstruction

Correct incipient fault diagnosis is crucial to the health management of the analog circuits, though remaining challenging. This paper presents a novel fault diagnosis method to diagnose ordinary soft fault and incipient soft fault for linear analog circuit. Due to the presence of analog circuit noise stress and the tolerance of the components, the linear circuit responses are considered as a stochastic process; and the time responses are acquired by sampling the outputs of the circuits under test. In this study, the local mean decomposition (LMD) method is used to decompose the stochastic series into a set of functions, each of which is the product of an envelope signal and a frequency modulated signal. The noises are decomposed and extracted from the sampled signals, which are later reconstructed. In the next step, the fault signatures are extracted from the de-noised series by applying the improved wavelet transform. Then the de-noised series samples are randomly selected as the training sample group and testing sample group. Finally, the feasibility of the proposed method is tested by SVM classifier. The proposed incipient fault diagnosis procedure is performed and verified on a Sallen-Key band-pass filter circuit and a four-op-amp high-pass filter circuit, which reveals improved accuracy of the proposed approach.