The Research On Fault Diagnosis Of Nonlinear Analog Circuits Based On Information Fusion

With the development of modern circuit theory, the fault diagnosis of analog circuits is becoming more and more concerned by scholars. It has become the network theory in the Third branches of the recognized only a short time from the beginning of the last century 60’s to now. At the same time, the mainly analog circuit signals are nonlinear, it is the reason that nonlinear analog circuit fault diagnosis become an extremely important part in fault diagnosis theory. So far,there has been a lot of research on fault diagnosis methods for nonlinear analog circuits.However, with the continuous development of the integration technology,it is clear that the application circuit has entered the era of high integration, the size of the circuit system is getting bigger and bigger, but the node is less and less, so that some features of the circuit fault are similar.In this paper, a fault diagnosis method based on double Wiener kernel is proposed to solve the problem that the single source feature is not sufficient. The method firstly uses the Gauss white noise as the excitation. After sampling, the circuit and the power supply of the two Wiener kernels are obtained by using the discrete Wiener kernel acquisition method. Then based on Improved Particle Swarm Optimization annealing, organic combination of feature selection and fusion as an optimization problem with the total Euclidean distance as the objective function. At last, this paper proposes a method of information fusion based on the total Euclidean distance weight distribution, which makes the different information organic fusion, and uses BP neural network to complete intelligent diagnosis.At the same time, this paper also designed the analog circuit fault diagnosis system based on ATmega128 single chip microcomputer. The hardware part mainly includes two parts of Gauss white noise generation and data diagnosis.And the design of the application in the diagnosis of PC software, which can achieve the acquisition circuit and the power of the Wiener kernel, obtain feature selection and optimal fusion of two Wiener cores, and train neural network learning and other functions. The fault circuit are measured by the system have been designed. The experiments show that can be more intuitive to diagnosis fault type and fault element. It is proved that based on Wiener kernel information fusion nonlinear analog circuit fault diagnosis method is feasibility. By comparing with the experimental results of single circuit Wiener kernel, the fault diagnosis method of nonlinear analog circuits based on Wiener kernel information fusion is more effective to distinguish the fault feature.