Study on Life Prediction Method of MOSFET Thermal Environment Experiments Based on Extended Kalman Filter

The prediction method of the residual service life of power MOSFET is studied in this paper. By analyzing the data collected under the existing thermal overload accelerated aging experiment, after processing the data, the failure threshold was set by using the prediction algorithm based on data drive and model to predict the residual life of power MOSFET. The traditional SVR algorithm requires a lot of parameter selection and only a few parameter convergence. The prediction algorithm model is based on Extended Kalman Filter, the extended Kalman filter is relative to the advantage of support vector machine (SVM) regression is used to predict the variance is small, and can be found in a wide range of required to predict the sample interval and the predicted results are more accurate, the test results verify the feasibility of this method.