Research On Fault Diagnosis Of Analog Circuits Based On Test Point Selection And Deep Features Coupling

With the development of electronic technology,electronic circuits are utilized extensively across a range of industries.Analog circuits are an significant part of electronic circuits,so the analog circuit fault diagnosis is crucial for the upkeep of electronic circuit systems.When analog circuit fault arises,timely locating and repairing the fault can make the circuit function properly.However,the complexity of diagnosing analog circuit faults has significantly grown due to the continuity,tolerance,and nonlinearity of analog circuit component parameters.This paper optimizes analog circuit fault diagnosis technology from analog circuit test point selection technology and analog circuit fault diagnosis technology.The details are as follows:(1)A test point selection method based on directed graph and isolation degree is proposed.In this paper,the circuit under tested is transformed into a directed graph according to the signal direction,and the importance of each test point in the directed graph is calculated by pagerank algorithm,and the similarity of each test point pair is calculated by simrank algorithm.The fitness function of genetic algorithm is constructed by the importance and similarity,and the test point set is preliminarily selected by genetic algorithm.The fault isolation degree of each test point in the test point set is calculated,and the inclusion-exclusion method ultimately chooses the optimal test point set.The method is not necessary to collect all the information of test points,and can initially filter the set of test points by using the directed graph.Finally,the inclusion-exclusion algorithm is used to select the optimal test point set,which eliminates the redundant test points.Simulation results show the effectiveness of the proposed method.(2)An analog circuit fault diagnosis method based on End-to-End Mutually Exclusive Autoencoder(EEMEAE)is proposed.The fault feature in analog circuits are very similar,which reduces accuracy of fault diagnosis.In order to make full use of the advantages of Fourier transform(FT)and wavelet packet transform(WPT)for extracting signal features,the original signals processed by FT and WPT are fed into two autoencoders respectively.The hidden layers of the autoencoders are mutually exclusive by Euclidean distance restriction.And the reconstruction layer is replaced by a softmax layer and 1-norm combined with cross-entropy that can effectively enhance the discriminability of features.Then concatenate the features of the two autoencoders to achieve deep feature coupling.Finally,the learning rate is adjusted adaptively by the difference of loss function to further accelerate convergence and diagnostic performance of the model.The proposed method is verified by the simulation circuit and actual circuit and the experimental results illustrate that it is effective.