Soft fault diagnosis based on the Pearson correlation

Abstract

The paper deals with the diagnostic algorithm enabling location and identification of soft faults in analogue circuits. It bases on the spectrum analysis of the circuit response to the rectangular input signal. The classifier applied in the location process uses Pearson product-moment correlation coefficient of the examined circuit with the models of all potential faults. The identification is based on the formulas constructed with use of one of the evolutionary methods: gene expression programming. The algorithm represents SBT (Simulation Before Test) technique and requires multiple analysis of circuit under test. They enable us to build a fault dictionary. The numerical example shows the effectiveness of the proposed algorithm.