Research On Prediction Method Of Remaining Useful Life And Thermal Fatigue Model Of Power MOSFET

Power MOSFETs are widely used as semiconductor components in people's daily life.Power MOSFETs play a vital role in various circuit systems.However,after a long period of use,the power MOSFET will eventually fail due to aging.In order to avoid paralysis of the entire circuit system and timely replacement of new devices,the remaining useful life(RUL)of the power MOSFET needs to be predicted.Therefore,based on the degradation data,this paper establishes a degradation model of power MOSFETs,and proposes to predict the remaining service life of power MOSFETs by extended kalman filter algorithm and improving particle filter algorithm.In addition,the finite element simulation software is used to build a thermal fatigue model of the power MOSFET and predict its service life.The whole research has certain guiding significance in the life prediction method of power MOSFET.First of all,this paper summarizes the methods and research work that domestic and foreign scholars have put forward to predict the service life of power semiconductor devices.Introduce the accelerated thermal aging test of the power MOSFET conducted by the NASA Research Center of the United States and process the test data,and select the appropriate test data as the research object of remaining life prediction.Then,the principles of the kalman filter algorithm and the extended kalman filter algorithm are outlined and the advantages and disadvantages of the two algorithms are compared.Due to the nonlinearity of the thermal aging test data of the power MOSFET,the extended kalman filter algorithm is used to predict the remaining service life,but the prediction effect is not ideal.Therefore,particle filter algorithm is used to achieve power MOSFET life prediction.Studies have shown that,compared with the prediction results of the extended kalman filter algorithm,the relative accuracy of the particle filter algorithm has been improved to some extent,but the prediction results are unstable.Therefore,this paper proposes to use an improved particle filtering algorithm — the unscented kalman particle filtering algorithm to predict the remaining life of the power MOSFET.The research shows that the relative accuracy of the proposed improved particle filter algorithm remains above95%,and the prediction accuracy is high and stable,which is suitable for the life prediction of power devices.Finally,based on the finite element method,a three-dimensional finite element simulation model of the power MOSFET with the package form TO-220 was established in the COMSOL Multiphysics simulation software.Through steady-state thermal analysis and transient thermal stress analysis of the model,the temperature distribution and thermal stress distribution of the power MOSFET during operation are obtained.The temperature curve ofthe chip with time is loaded into the fatigue model as a thermal cycle load,and the service life of the model is predicted according to the Coffin-Manson fatigue failure criterion.