Analog Circuit Fault Diagnosis Based On Best Test Points Selection And Improved MRMR-SVDD

With the development of science and technology,electronic circuits have achieved great progress.A variety of electronic equipments have become an indispensable part of human society.Analog circuit is the basis of electronic circuits,once the analog circuit failure,the entire circuit will not operate correctly.To be more serious,those failure may cause disasters.The fault diagnosis and prevention of analog circuits are the keys to ensure the normal operation of the whole electronic circuit system,which is of great significance to improve the performance of electronic equipment.In this paper,the analog circuit diagnosis technology is optimized from the three aspects of test points optimization,optimal feature subset extraction and fault pattern classification,:(1)Optimize the test points selection model based on gray relational analysisDue to the limited number of simulated circuit fault samples,the fault information is not rich enough.The analog circuit has the characteristics of gray system,and the analog circuit system can be regarded as gray system.Based on the introduction gray relational analysis model,we analyze the shortcomings of the model and improve it.Gray entropy is used as a measure of the sensitivity of different test points,and a fault detection point selection technique is proposed to modify the gray relational analysis.(2)Equilibrium weighting factor mRMR principle optimal fault feature selection modelFirstly,we give a simple introduction about Wiener series,and then the method of solving low-order Wiener kernel is introduced.Secondly,we introduce the mRMR principle,put forward its shortcomings and improve it.Finally,a fault feature selection model based on improved mRMR principle is proposed.(3)Adaptive gradient algorithm combined kernel SVDD fault classification modelThe combination of the Gaussian kernel function and the polynomial kernel function is the new SVDD kernel function.The new kernel function has better generalization ability and classification accuracy.In order to finding the weight coefficient of the combined kernel function,the improved gradient algorithm is used to solve the local maximization problem of the gradient algorithm at the extreme points.