A Research On Fault Diagnosis Based On Neural Network In Analog Circuit

Fault diagnosis in analog circuits is a comparatively front research topic. Progress in very deep submicron semiconductor technology drives the advent of System-on-Chip (SOC) and analog/digital mixed-signal integrated circuits. Many new theoretical problems appear in the analog test area. Using normal or traditional theories and methods of fault diagnosis, they are difficult to be solved. For the neural network has the function of processing complex multi- patterns and carrying on the association, the extrapolation and the memory,it is particularly suitable for fault diagnosis system. Fault diagnosis by applying the neural network in analog circuit has become the most recent development tendency。This paper studies the extraction of fault features and the method of fault diagnosis in analog circuit by applying neural networks in fault diagnosis, in light of Fuzzy theory, wavelet transform theory and the theory of information fusion.This paper introduces the basic concepts, the methods of neural network and so on, and focuses on its structure and the method of choosing network parameter. Using the BP algorithm and the strong computational capabilities and high accuracy of DSP, the analog circuit fault diagnosis system is designed, which is more accurate, convenient, and has a higher practical value than traditional Fault Dictionary Method. The paper, basing on the theory and methods of fault diagnosis and applying the theory of artificial intelligent, fuzzy theory and neural network, proposes the method of integrated fuzzy reasoning and neural network, introduces the structure and the working principles of comprehensive system. A fault diannostic method based on interval value fuzzy neural networks for condition monitorinn and fault diannosis of in lame machinery is proposed.The paper elaborates the basic theory of wavelet analysis and wavelet packet analysis, discusses the general framework of the multi- resolutions analysis, gives the definition and the nature of Wavelet analysis, as well as, from the perspective of space decomposition, understands wavelet packet analysis and proposes the algorithm of wavelet packet decomposition and reconstruction. It proposes that fault diagnosis in analog circuits based on the Wavelet - Neural Networks extracts fault feature of analog circuits by applying multi- resolutions analysis and Wavelet Transform. Then it offers one detailed fault diagnosis process. For the diagnosis method based on Neural Network Information Fusion Technology belongs to the typical characteristic level fusion, it introduces the diagnosis model of neural network information fusion and finally prove the feasibility of this method through the simulation experiment on the international standard electric circuit.