Research On Incipient Fault Diagnosis Method Of Analog Circuit Based On Test Data

With the rapid development of electronic technology,the electronic system is becoming larger and more complex in structure.Analog circuit plays an irreplaceable role in the electronic system.With the wide application of the electronic system,the research on fault diagnosis of analog circuit has been paid much attention by scholars at home and abroad.However,most of the current fault diagnosis methods for analog circuits are mainly aimed at hard faults or severe soft faults,and there is little research on incipient fault diagnosis of analog circuits.With the increase of service time,the components of analog circuits degrade to a certain extent inevitably.At this time,the performance of analog circuits will also be affected but not completely invalidated.This state is called incipient fault of analog circuits.If the incipient fault of analog circuit is allowed,it will inevitably lead to the complete failure of electronic system.Therefore,this thesis takes the incipient fault diagnosis of analog circuit as the research topic,focusing on the feature extraction and fault mode classification method of analog circuit incipient fault.The main work of this thesis is as follows:The definition of incipient fault of analog circuit is defined based on the mechanism research of analog circuit incipient fault.This thesis analyzes the causes of the incipient fault in analog circuit,clarifies that the degradation of the circuit element is the main cause of the incipient fault of the analog circuit,and expounds the degradation laws of several common circuit elements.The sensitivity and fault propagation characteristics of analog circuits are studied as a preparation for the research of incipient fault diagnosis methods of analog circuits.Aiming at the fact that the incipient fault features of analog circuits are not obvious,an incipient fault feature extraction method for analog circuits based on improved empirical wavelet transform(IEWT)and supervised locally linear embedding(SLLE)algorithm is proposed.The initial eigenvectors are composed of energy entropy and kurtosis of IEWT component signals of circuit test point signal,and then the eigenvectors are reduced by using the SLLE algorithm.After that the eigenvector sequence of each incipient fault mode of analog circuit is obtained,which solves the problem of the overlapping of different incipient fault modes in analog circuit.Aiming at the problem of incipient fault location in analog circuit,this thesis presents a method of incipient fault mode classification of analog circuit based on hidden Markov model(HMM)optimized by artificial fish swarm algorithm(AFSA).This method trains several independent HMMs as the pattern classifier for incipient fault of analog circuit.According to the defects of the traditional HMM training algorithm,the AFSA is used to optimize the model parameters of HMM,and the fault location of analog circuit incipient fault is realized.The effectiveness of this method is illustrated by simulation circuit experiment.In order to verify the effectiveness and superiority of the method in this thesis,the actual analog circuit is built,and different incipient fault modes are simulated by adjusting the resistance of rheostat or replacing capacitors with different capacitance values.The incipient fault diagnosis method of analog circuit based on IEWT-SLLE and AFSA-HMM proposed in this thesis is applied to the incipient fault diagnosis of actual analog circuit,and the validity of the methods in this thesis is verified.The superiority of the methods in this thesis is further illustrated by the comparative experimental analysis.