Research On Fault Detection And Fault Diagnosis Of Incipient Fault In Analog Circuits Based On Compactness Measuring

With the rapid development of electronic technology,electronic systems are widely used in many important fields,such as aerospace,national defense,transportation and medical treatments.Therefore,the reliability of electronic systems has brought to attention,and many scholars have deeply studied the diagnosis methods for analog circuits.Fault states in analog circuits are composed of hard fault and soft fault,and these days,studies are focusing on hard fault and soft fault with bigger deviation.There are few study on incipient fault,which is the early stage of soft fault.When circuit is under incipient fault state,the performance is getting worse,however,with little fault feature and circuit tolerance,there is great overlap between normal state and incipient fault state.Therefore,traditional fault diagnosis methods cannot achieve pleasing results,and incipient fault detection and diagnosis have become a challenging issue.To solve the above problem,this paper takes incipient fault in analog circuits as studying subject,deeply studied diagnosis methods for incipient fault in analog circuits.First,a testing stimulus generation method based on compactness measuring algorithm has been proposed,which can stimulate the fault feature in analog circuit under test to the greatest degree.Then the wavelet analysis method has been used for fault feature extraction in analog circuits,when using the previous generated stimulus.In fault detection step,two incipient fault detection methods based on support vectors have been proposed: incipient fault detection method based on support vector machine and incipient fault detection method based on support vector data description.The detection results of the above two methods have been compared.For fault diagnosis step,two fault diagnosis methods based on machine learning have been proposed: fault diagnosis method based on multi-classification support vector machine and fault diagnosis method based on deep neural network.To verify the proposed methods’ practical effect,stimulation experiments and hardware experiments have been conducted in Sallen-Key filter,CTSV filter and Leapfrog filter,which all belong to ISCAS’ 97.Also,the detection and diagnosis results have been compared with results from other methods.The above results indicate that the proposed methods in this paper have better effect in incipient fault detection and diagnosis in analog circuits.