A method for minimum node selection in diagnostics of analog systems

Adrian Bilski, Jacek Wojciechowski

Abstract

The aim of this work is to introduce a strategy for finding minimal set of test nodes for diagnostics of complex analog systems with single parametric faults, using SVM classifier as a fault locator. The results of diagnostics of a low-pass filter using taboo search and GA algorithms as node selectors in conjunction with SVM fault classifier are presented.

Keywords

complex analog systems, support vector machine, taboo search, genetic algorithm, parametric fault detection

Download

IMEKO-TC10-2014-014.pdf

DOI

-

IMEKO TC

TC10 - Measurement for Diagnostics, Optimization and Control

Back to the proceedings