Fault Diagnosis Method Of Analogue Integrated Circuits Based On Wavelet Theory And LSSVM

With the development of the micro-electronics technique, it is urgent to fault diagnosisfor the integrated circuit.The faults of micro-electronics above90%come from theintegrated circuit.With the rapid development of the integrated circuit technique, the scale ofelectric circuit is bigger and biger,and diagnostic difficulty is higher and higher.The researchof fault diagnosis method of analogue integrated circuits has already become the importantresearch topic in the circuits diagnosis field. Fault diagnosis method of analogue integratedcircuits is performed to realize the automation and intelligence of fault diagnosis ofanalogue integrated circuits,which has a very high utility value.Because of diversity andcomplexity of the failures of analogue integrated circuits, traditional fault diagnosis methodis difficult to the expected diagnosis effects.Thus, fault diagnosis method of analogueintegrated circuits based on wavelet theory and least squares support vectormachine(LSSVM) is presented.Firstly,the structure of fault diagnosis system of analogueintegrated circuits based on wavelet theory and LSSVM is proposed in the thesis.Diagnostic characteristics extraction method of analogue integrated circuits based onwavelet theory is presented,and fault recognition of analogue integrated circuits based onLSSVM is presented.Finally, fault diagnosis system of analogue integrated circuits based onwavelet theory and LSSVM is developed.Main research work and achievement of thesis are given as followings:1) In order to solve the problems of the low precision of feature extraction and faultdiagnosis in the fault diagnosis for analogue integrated circuits,the structure of faultdiagnosis system of analogue integrated circuits based on wavelet theory and LSSVM ispresented.Firstly, fault characteristics of analogue integrated circuits are analyzed, thestructure of fault diagnosis system of analogue integrated circuits based on wavelet theoryand LSSVM is created according to the fault characteristics of analogue integrated circuits.2) Diagnostic characteristics extraction method of analogue integrated circuits based onwavelet theory is presented. Firstly, wavelet theory is introduced. Then,diagnosticcharacteristics extraction model of analogue integrated circuits based on wavelet theory isestablished. The case is used to test the diagnostic characteristics extraction method.3) Fault recognition of analogue integrated circuits based on LSSVM is presented.Firstly, least squares support vector machine classifier is introduced,and parameteroptimization of LSSVM based on chaos particle swarm optimization algorithm is presented,fault recognition model of analogue integrated circuits based on LSSVM is presented,and the case is analyzed.4) In the basis of key technologies of fault diagnosis for analog integrated circuitsbased on wavelet theory and least squares support vector machine,fault diagnosis system ofanalogue integrated circuits based on wavelet theory and LSSVM is developed. Firstly,overall structure and workflow of fault diagnosis system of analogue integrated circuitsbased on wavelet theory and LSSVM is given,the functional model of the system isintroduced, fault diagnosis model of analogue integrated circuits based on wavelet theoryand LSSVM is realized.Finally,the cases are used to test the model.The experimental resultsshow that the diagnosis system is effective.