基于小波包分析神经网络的模拟电路故障诊断方法研究模式识别与智能系统专业论文.docxVIP

Study on Fault Diagnosis in Analog Circuits Based on Wavelet Packet Analysis and Neural networks ABSTRACT Because of the analog circuit’s inherent particularity, the development of it’s fault diagnosis technology is relatively slower than digital circuit, and the research field of the analog circuit fault diagnosis is all along the challenge and hot focus in electrical circuit system’s fault diagnosis. Along with the development of microelectronics technology, integrated circuits have been largely used into circuit systems, and some new methods of diagnosis which are superior to the traditional methods have been applied in hybrid circuits. Neural Network and its optimized algorithms have been introduced into fault diagnosis fields, and created a new situation for the study of the analog circuits’ fault diagnosis. Wavelet packet analysis can adaptively determine the signals’ resolutions in different frequency bands, and signals’ synchronized segregation from the high frequency to low can be realized; Neural Network’s advantages are its strong nonlinear mapping, self-learning, adaptive and fault tolerant abilities and so on. In view of both advantages, this paper combines the wavelet packet and neural network algorithm to deal with the problems in analog circuit fault diagnosis field and makes some simulation analysis to the circuit examples, which meets the desired diagnostic targets. The main tasks of this paper are as follows: Firstly, on the basis of the analysis and comparison of the commonly used methods applied to the analog circuit fault diagnosis, this paper details the general theories of Neural Network, wavelet and wavelet packet analysis, and completes Neural Network structure’s design which is based on wavelet packet. Secondly, through an example of circuit, a diagnostic method of “energy-fault” which is based on wavelet packet is studied and applied into the fault information extraction for analog circuit, and thus to simplify network