Research On The Reliabiabity Evaluation For BUX10Based On Performance Degradation Data

With the advancement of science and technology，the demand for the reliability ofproducts is increasingly high. In traditional reliability evaluation methods, a lot of life dataare required, and the experimental period is long. It is not fit for the present requirements.The reliability can be assessed based on the relationship between failure mechanisms anddegradation processes. This method doesn’t need failure data, and its experimental periodis short. It is an effective method of solving the reliability problems.This paper mainly conducts the degradation experiment on the high-power transistors,named BUX10, made by ST. Then the thesis designs the degradation experiment programfor BUX10, and analyses their reliability through degradation trajectory models. It alsoprovides a way to predict the remaining life of BUX10based on the failure data. Thisthesis uses the Bayes method to get more precise analysis results, and analyzes thereliability of BUX10by means of the actual testing data. The main research workincludes:(1) Research common analysis methods of reliability based on the performancedegradation, analyze relative theories of the reliability degradation, and summarize themain steps towards analyzing reliability.(2) Research structural features and the failure mechanism of BUX10, design twodegradation experiment programs, i.e. long trial testing program and accelerateddegradation testing program, and deeply analyze the long trial testing program.(3) Use the software of Matlab2007to analyze the reliability of high-powertransistors named BUX10according to the data obtained from the long trial testing.(4)Research the Bayes theory and use the Bayes method to analyze the newdegradation data and obtain the updated reliability characteristic values.The purpose of this thesis is to estimate the failure lifetime of BUX10and otherreliability characteristic values by researching the high-power transistors that have thelong lifetime and need high testing cost. So it provides an effective way for the reliabilityassessment of electronic components with small samples. Moreover, the results havesignificance in the practical projects.