Parametric Fault Diagnosis of Analog Circuits using Adaptive Boosting

This work proposes an improved parametric fault diagnosis technique targeting increased analog fault coverage regardless of the components to be tested. Exciting the available circuit input node with a pseudo-random signal and fault diagnosis using machine-learning algorithms by analyzing its output response is presented in this work. Moreover, a classifier, termed Adapting Boosting (AdaBoost), is developed by combining two different training algorithms, namely artificial neural network (ANN) and Support Vector Machine (SVM), along with the Probabilistic Principal Component Analysis (PPCA) and Wavelet Transform, is applied for the fault diagnosis of analog circuits. The proposed algorithm is validated and its effectiveness is shown for two benchmark circuit i.e., second-order Sallen-Key band-pass filter.