Analog Circuit Fault Diagnosis Algorithm Based on KPCA Feature Optimization Combined with Improved GA - SVM

To enhance the accuracy of analog circuit fault diagnosis, this paper presents that kernel principal component analysis can be used to optimize the analog circuit fault feature vector extracted by the time domain statistical analysis, eliminate the redundant information of the fault feature vector, and finally constructs the optimal fault feature vector set. Aiming at the shortcoming that basic genetic algorithmis easy to trap into local extremum, this paper also improves the standard genetic algorithm byintroducing fertility group synergy evolution and adaptive mutation probability. According to what are mentioned above, this paper presents the method of analog circuit fault diagnosis based on KPCA feature optimization combined with improved GA-SVM. Simulations demonstrate that the proposed approach achieves fault diagnosis accuracies of 100<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup> for the sallen-key circuit and 94% for the chebyshev filter circuit, representing improvements of 11 % and 9% over traditional SVM methods.