Analog Circuit Fault Diagnosis Based On Wavelet Analysis And Genetic Neural Network

Analog circuit testing and fault diagnosis has been a hot studied project since the1960s,but analog circuit fault diagnosis has become somewhat difficult because thetype of fault is more complex,and component parameters in the circuit has a greattolerance,and susceptible to outside interference,and non-linear components in thenetwork nonlinear problems,and test equipment is not always ideal output and soon.Neural network which is represented by artificial intelligence,as an adaptivepattern recognition technology,through its own learning mechanism automaticallyform required by the decision-making area,provides an effective approach to solveproblems of fault diagnosis of analog circuits,so it is given greatly attention byacademic circles.In this paper,a general analog circuit fault diagnosis methods and diagnosticprocedures are discussed,the structure and training algorithm of BP neural networkare studied,and the advantages of the neural network used to analog circuit faultdiagnosis are introdeced,which combined with the latest achievements of the neuralnetwork technology.The genetic algorithm can overcome the BP network’sshortcomings that are not easy to converge and easy to fall into the local minimum,wavelet analysis can effectively deal with the input data and extract the fault signalfeature vector.So a new method of analog circuit fault diagnosis based on geneticalgorithms and wavelet neural network is proposed.Throughout this paper,the main line is neural network,and the research object ismultifunction filter circuit,combined wavelet transform and genetic algorithm toapply to the analog circuit fault diagnosis.Simulate the fault diagnosis process usingORCAD and MATLAB software,In the diagnostic process on the analog circuit,usingwavelet decomposition as a preprocessor,thoughout the principal componentanalysis,effectively reducing the input dimension of neural networks;And then usegenetic algorithm to optimize the BP neural network’s weights and thresholds,toovercome the shortcomings of the BP network,to improve the structure of BP neuralnetwork.The results show that wavelet analysis can effectively deal with the inputdata and extract the fault signal feature vector,shorten the time of the fault diagnosis,by joining the genetic algorithm,the BP neural network structure is greatly improved,the number of training steps is reduced,the ability of the recognition is improved.This method provides an effective method for fault location.