Neural Network based Fault Diagnosis in Analog Electronic Circuit using Polynomial Curve Fitting

Abstract

Many studies have been presented for the fault diagnosis of electronic analog circuits with worst case fault models using ±50% variation in the parametric values of the components. The study of for parametric fault detection in electronic analog circuits-faults as small as 10% or less was uncovered. The use of the neural network for parametric fault diagnosis in an analog circuit, based upon the polynomial curve fitting coefficients of the output response of an analog circuit is presented in this study. Building upon the theory of polynomial coefficients we propose a parametric fault diagnosis methodology. A polynomial of suitable degree is fitted to the output frequency response of an analog circuit. The coefficients of the polynomial attain different values under faulty and non faulty conditions. Using these features of polynomial coefficients, a BPNN is used to detect the parametric faults. Simulation results are presented for a benchmark bi quad filter circuit. Single resistance and capacitance faults of ±1% to ±50% deviation from nominal values were correctly diagnosed.