Diagnosis Of Analog Circut Fault Based On Wavelet And Mahalanobis Distance

Diagnosis of analog circuits is constraining the development of modern electronic technology.It’s very important to find a way to carry out accurate and reliable method of analog circuit faultdiagnosis. This paper is committed to reach a method of analog circuit fault diagnosis, which not onlydetermine whether there is a fault, but also does the successful positioning of faulty components in theanalog circuits.Wavelet transform which has the great ability to display details of singals, is used as signalpre-processing. On the other hand, Mahalanobis distance is good at calculating the sample similarity.Based on wavelet transform and Mahalanobis distance, a method of analog circuit diagnosis withtolerance is proposed in this paper. Main work of this paper is as follows:1. A program of analog circuit diagnosis with tolerance is designed based on the theory of theclassical analog circuit fault diagnosis, the wavelet analysis and the Mahalanobis distance.2. Reach of the signal pre-processing based on wavelet analysis and Mahalanobis distance-baseddiagnosis of the specific implementation is done. And some conclusions are made after theoreticalanalysis and repeated experiments.3. Aiming at the single catastrophic and the single parametric fault, examinations of the proposedmethod of analog circuit aiagnosis with tolerance are done in PSpice and Matlab.After testing, this method based on wavelet analysis and Mahalanobis distance which proposedin this paper has a great positioning ability in analog circuit diagnosis with tolerance, the highaccuracy of single catastrophic or the single parametric fault detection and the advantage of simplicityand small online computational amount.