Study On Reliability Evaluation Technique Of Aluminum Electrolytic Capacitor In SMPS

The reliability of the switching mode power supply(SMPS)has an important impact on the stable operation of the avionics system.More than 60% of the SMPS failures are caused by the failure of the output electrolytic capacitor.The traditional offline electrolytic capacitor monitoring method needs to remove the capacitor from the system,which interrupts the continuous work of the system.Therefore,an effective electrolytic capacitor online monitoring method is urgently needed.In this thesis,the Buck circuit is taken as an example of SMPS.Based on the analysis of the failure mechanism of the aluminum electrolytic capacitor,the equivalent series resistance(ESR)of the aluminum electrolytic capacitor is selected as a characteristic parameter for the degradation of the capacitor.An online monitoring method of output capacitor without current sensor is proposed.Whether the circuit is in continuous conduction mode(CCM)or discontinuous conduction mode(DCM),it is convenient to monitor the capacitor.The main works of the thesis are as follows:An on-line monitoring method for the output capacitor's ESR of the Buck circuit based on the trigger mode is studied.This method analyzes the voltage and current expressions of the circuit at CCM and DCM to establish the output ripple voltage expression.The ESR monitoring model of the output capacitor is established by the output ripple voltage at time 0 and DT.Further,in view of the shortcomings such as poor robustness of this model at DCM,the ripple voltage at time /2,3 /4 and DT is used to establish an ESR monitoring model of the output capacitor,which eliminates the effect of the equivalent series inductance(ESL)and enhances the model robustness.Simulation experiments show that the average relative error of ESR monitoring in this method is 1.4% at CCM and 0.4% at DCM.The output ripple voltage of the Buck circuit has the characteristics of Time-sequence.A data set was established by collecting ripple voltage data of capacitors with different aging states,and an on-line monitoring model of output capacitor's ESR based on bidirectional Long Short-Term Memory(LSTM)neural network was established.This method does not require the use of a trigger circuit,simulation experiments show that the average relative error of ESR monitoring in this method is 4.6% at CCM and 4.1% at DCM.In order to verify the effectiveness of the proposed methods,an open-loop Buck circuit with controllable switching frequency and duty cycle was adopted as the target circuit.On-line monitoring experiments of output capacitor were developed in two modes of CCM and DCM under different switching frequencies,different loads and different temperatures.Capacitor aging test was also carried out.The monitoring results of capacitor's ESR were compared with the test results of LCR testers.The experimental results show that the output capacitor's ESR on-line monitoring improved method based on trigger mode has an average relative error of 1.8% at CCM and 1.0% at DCM.The output capacitor's ESR on-line monitoring method based on the bidirectional LSTM neural network has an average relative error of 4.2% at CCM and 3.6% at DCM.The above experimental results prove that the method proposed in this paper effectively solves the online condition monitoring problem of aluminum electrolytic capacitors in practical applications.