Arvind Sai Sarathi Vasana, Bing Longa,b and Michael Pechta,c
aCenter for Advanced Life Cycle Engineering (CALCE), University of Maryland, College Park, MD-20742, USA
bSchool of Automation Engineering, University of Electronic Science and Technology of China, Chengdu, China
cCenter for Prognostics and System Health Management, City University of Hong Kong, Kowloon, Hong Kong
Analog circuits have been widely used in diverse fields such as avionics, telecommunications, healthcare, and more. Detection and identification of soft faults in analog circuits subjected to component variation within standard tolerance range is critical for the development of reliable electronic systems, and thus forms the primary focus of this paper. In this paper, we have experimentally demonstrated a reliable and accurate (99%) fault diagnostic framework consisting of a sweep signal generator, spectral estimator and a least squares-support vector machine. The proposed method is completely automated and can be extended for testing other analog circuits whose performances are mainly determined by their frequency characteristics.
Keywords – analog circuits, frequency features, least squares-support vector machines, soft fault diagnosis, tolerance, wavelet features.
Full text available to CALCE consortium members.