Analog circuit fault diagnosis using bagging ensemble method with cross-validation

Abstract

Neural Network (NN) ensemble approach has been an appealing topic in the field of analog circuit fault diagnosis lately. In this paper, a new method for fault diagnosis of analog circuits with tolerance based on NN ensemble method with cross-validation is proposed. Firstly, bias-variance decomposition shows the theoretical guide on how to choose the component networks when composing the ensemble. Secondly, output voltage signal of the Circuit Under Test (CUT) has been obtained after the stimulus imposed on the CUT. After getting the corresponding fault feature sets, Bagging algorithm is employed to produce the different training sets in order to train the different component networks, and cross-validation technique has been employed to further improve fault diagnosis accuracy. Finally, the outputs of the component ensemble members are combined to isolate the CUT faults. Simulations result shows the superior performance of this proposed approach. This system is able to effectively improve the generalization ability of the analog circuit fault classifier and increase the fault diagnosis accuracy.