Research On Health Prediction Method Of Analog Circuits Based On Empirical Wavelet Transform

With the rapid development of electronic technology,the analog circuit structure is becoming more and more complicated,the incidence of circuit failure is getting higher and higher.As the analog circuit is running,the fault performance is nonlinear and uncertain,It is difficult to extract the fault characteristics of the circuit accurately and effectively for the conventional fault feature extraction method.It is an inevitable trend to use new method to extract the fault characteristics of the analog circuit.The empirical wavelet transform(EWT)is a new adaptive signal analysis method developed in recent years.It is mainly used to deal with non-stationary and uncertain random signals.In order to obtain the optimal fault characteristic of the analog circuit.this paper applies the EWT method to the fault feature extraction of the analog circuit,and proposes a new method of fault feature extraction based on empirical wavelet transform.In order to verify the method,this paper has carried on the experiment simulation through the specific circuit,and extracts the fundamental frequency component of the fault signal under different fault modes of the circuit.Combined with the energy entropy of the signal,the failure characteristic of the circuit is successfully extracted.the simulation results show the feasibility and effectiveness of the proposed method.At present,the research has been very rich at home and abroad on the analog circuit fault diagnosis,but the reports are relatively rare on the analog circuit health prediction.The fault feature data extracted by the existing fault prediction method can not reflect the degradation degree of the circuit components well and can not effectively predict the health status of the analog circuit accurately.In this paper,based on the empirical wavelet transform method,the concept of circuit health index(HI),which characterizes the degradation degree of circuit components,is proposed,and the quantitative relationship between the fault health index and the fault characteristics is studied.The evaluation mechanism of the health status of the analog circuit is established,combined with BP neural network prediction model,completed the simulation of the circuit.Finally,the experimental results show that the error prediction data of EWT are small compared with the physical data,which can predict the degradation degree of component fault accurately and are very suitable for practical analog circuit health prediction.