A novel fault diagnostic method for analog circuits using frequency response features

Abstract

Analog circuits are an important component of complex electrical systems. Therefore, fault diagnosis of analog circuits plays a vital role in ensuring the reliability of electronic systems. A novel fault diagnostic method for analog circuits based on the support vector machine (SVM) optimized by the firefly algorithm (FA) using frequency response features is presented in this paper. Wilks Λ-statistic can effectively assess the ability of variables to resolve multiple types of samples in multivariate statistical analysis. Frequency responses of analog circuits are measured, and then, features are extracted by using the particle swarm optimization (PSO) method. Additionally, the fitness function of the PSO is set to Wilks Λ-statistic. Then, an SVM based analog circuit’s fault diagnosis model is introduced to classify the faulty components according to the extracted frequency response features. The optimal penalty parameter and kernel function parameter of SVM are obtained by using the FA. The method is fully evaluated in fault diagnosis simulations of the Sallen-Key bandpass filter and four-op-amp biquad high-pass filter. The experimental results demonstrate that the proposed fault diagnostic method can produce higher diagnosis accuracy than other typical analog circuit fault diagnosis methods.