Study On Fault Diagnosis For Analog Circuit Based On Wavelet Analysis And Neural Network

Though digital circuit has had a perfect development, it cannot completely replace the analog circuit. At present, most of the faults in electronic equipment are caused by analog circuits. The reliability of the system depends largely on the reliability of the analog circuit. So doing research on analog circuits becomes particularly important. As the wavelet analysis can express local signal feature both in time-domain and frequency-domain, and at the same time the neural network has the advantages of dealing with the complex modes, speculation, memory, etc. So using wavelet analysis and neural network for fault diagnosis of analog circuit is an inevitable trend.The main works of this paper are shown as follows:Firstly, subject background, research status and development directions are introduced. Theories of wavelet analysis, including multi-resolution analysis, fast algorithm of wavelet, wavelet packet definition and nature are described. Besides, the neural network structure and learning rules are discussed too.Secondly, two integrating manners between wavelet transform and neural network are discussed. And then the basic principle and arithmetic of these two kinds of wavelet neural network are introduced. The simulation result shows that these integrating two manners are both effective in fault diagnosis of analog circuits.Thirdly, as the multi-wavelet transform has more advantages than the scalar wavelet transform, a method of fault diagnosis in analog circuits based on multi-wavelet neural network is proposed. Through using multi-wavelet transform for the data pretreatment, the structure of the neural network is simplified, and the training speed is accelerated. The simulation results verified that this proposed method is effective for the fault diagnosis of analog circuits and has better performance than WNN.Finally, a method for fault diagnosis of analog circuits based on the combination of wavelet packet analysis and RBF neural network is presented. The simulation result shows that the proposed method is effective in fault diagnosis of analog circuits.