Analog Circuit Fault Diagnosis Based On Wavelet Analysis And Neural Network

Analog circuit test and fault diagnosis have been a challenging subject in thefield of modern circuit, and have made many considerable work and methods so far.With the rapid development of the electronic industry, the importance of the analogcircuit fault diagnosis is more and more obvious. Unfortunatley, due to many factors,such as tolerance of components, diversity and comlexity of fault models and so on, itis difficult for the traditional fault diagnosis methods to achieve the expected resultsin practical applications.At present, intellingence information processing technologieslike Neural Networks, Wavelet Analysis, Swarm Intelligence Optimization and so on,have been a new hotspot and provided an effective way to slove the problem of analogcircuit fault diagnosis.This paper firstly reviews the research of analog circuit fault diagnosis, thenfocuses on the appication of wavelet transform and wavelet packet in analog circuitfault diagnosis aiming at the problems in accessing to fault information, and thesuperiority of the wavelet packet analysis is shown through the simulation example.For the Neural Networks have good learning ability and approximationperformance, it is widely used in analog circuit fault diagnosis. But the structure andmodel parameters of network are difficult to set. The paper introduces a swarmintelligence optimization algorithm-Praticle Swarm Optimization(PSO), and a newanalog circuit fault diagnosis method combined wavelt packet, Radial BasisFunction(RBF) network and PSO algorithm is proposed. The response signals ofanalog circuits are preprocessed by wavelte packet decomposition, and the optimalparameters of the network are obtained by the PSO algorithm. The method not onlyreduces the dimension of the input vector, but also avoids the blindness of parameterselection. Simulation results show that the proposed method is effective.In order to overcome the difficulties in feature extraction and fault classificationof analog circuit fault diagnosis, a new fault diagnosis method based on multiwaveletand Extreme Learning Machine(ELM) is proposed. Fault feature information areextracted by multiwavelet which has symmetry, orthogonality,short support and highorder vanish moments, then ELM is applied to classify the fault patterns. Thediagnostic example results show this method is effective and accurate for faultlocation of analog circuits.