A novel approach of analog circuit fault diagnosis utilizing RFT noise estimation

Abstract

This paper proposes a novel approach of analog circuit fault diagnosis utilizing Ramanujan-Fourier Transform (RFT) noise estimation and this method extracts circuit noise by performing RFT operation on the circuit output response. Diagnose whether the circuit has failed according to the noise standard deviation. This method is simpler than traditional Support Vector Machine, neural network, decision tree, Extreme Learning Machine and other artificial intelligence methods, without complicated algorithm, only need to perform cause-effect reverse reasoning to realize fault diagnosis, which can avoid a lot of sample training. It can be used as a new idea for analog circuit soft fault diagnosis. The correctness and feasibility of this approach are confirmed by experiments.