FAULT DIAGNOSIS IN ANALOG ELECTRONIC CIRCUITS - THE SVM APPROACH

Abstract

In this paper, the application of the SVM (Support Vector Machine) algorithm has been used for diagnosis and tests of analog electronic circuits. The diagnosis procedure belongs to simulation-before-test techniques, where simulations of the circuit under test (CUT) are performed at the before-test stage. Two examples have been verified for parametric and catastrophic faults in the time domain, but the conclusion is driven with the use of assumed features. A fault-driven test (FDT) has been applied to a filter circuit and a specification-driven test (SDT) to a field-programmable analog array (FPAA). The SVM classifies features which are calculated from the time domain responses. Results obtained from the examples prove a high detection and localization level of circuit states with the use of the SVM classifier.