Recent advances in symbolic techniques for analog fault diagnosis

In this work some advances in the use of symbolic methods for the fault diagnosis of analog circuits are presented. In particular a procedure for the selection of test frequencies in multifrequency parametric fault diagnosis is proposed. It is based on the evaluation of algebraic indices, as condition number and norm, of a sensitivity matrix of the circuit under test. During preliminary phase testability and ambiguity group concepts are used in order to evaluate the optimum set of testable components. By exploiting symbolic analysis techniques, a program implementing the proposed procedure has been developed. It permits, by means of an optimization procedure based on a genetic algorithm, to select the set of frequencies which better leads to locate parametric faults in analog circuits.