Methods And Realization For Fault Diagnosis Of Analogy Circuits Based On Wavelet Neural Network

With the development of electronic technology, the integration density of modern electronic circuit’s is becoming higher and higher. As the IC’s production technology has reached the Nano-grade, the size of electronic circuit’s is getting smaller and smaller. However, the vulnerable of electronic components haven’t improved, especially analog components. The stability, reliability, fault self-diagnosis performance of modern equipment’s running status is increasing, so that the technology of fault diagnosis get deeply research in the world.As an intelligent classifier, Neural network have many characteristics include nonlinear mapping, the ability to learn and adapt to environment. So, Neural network has been widely applied in fault diagnosis. As a graphic programming language, Lab VIEW is the core of the Virtual Instrument Technology, have lots of advantages such as the high efficiency in programming, very quick during operation and the development cycle is short. All of that Lab VIEW are widely used in fault diagnosis. At first, this paper is researched the structure, principle, design method of BP and wavelet Neural network, and designing approaches. Then, elaborated the theory of fault diagnosis based on neural network.This paper have detailed researched the application of wavelet analysis in extracting fault features, and discusses the algorithm theory, the concrete design step of multi-resolution analysis and energy analysis in fault diagnosis, propose the theory of joint failure features.Also, this paper have studied the principle of working, and derived the performance curves for nonlinear resistance on the Chua chaotic circuit. The circuit’s simulation instrument implement in the virtual instrument development platform.Finally, the fault diagnosis system based on wavelet neural network is implemented in the virtual instrument development platform. As the example of Chua circuits is diagnosed, make use of the different kinds of data processing method and different neural network model to verify the accuracy of the proposed method. It is very convenient to get the fault type and the fault location from virtual instrument front panels through human interface. So that the fault diagnosis based on virtual instrument technology is very practical value in the future.