Random Reliability Test Data Treat Method And Software

Method and software are investigated for the treat of random reliability test data. Following three kinds of work are performed.First, by the principle of censored reliability life test to instead of the parameter life with motioned test variable, the censored reliability test can be extrapolated to the wide fields of reliability tests. And the test and data-treat methods of censored reliability life test can be also extended to be applied widely in common reliability tests.Second, a new method is established for determining appropriate statistical model of the imperfectly failure test data, which consist of partial failure and partial non-failure, of censored reliability test. Three considerations are given. They are the goodness-of-fit, consistency with failure physics, and the safety prediction in tail region, similar to the previous work of Zhao. The difference is that the present work uses maximum likelihood method to determine the goodness-of-fit and validates to imperfectly failure test data, while Zhao et al's work uses the Pearson parameter-linear relationship coefficient to determine the goodness-of-fit and validates to perfect failure test data. Further, a framework for the treat of test data established. This framework consists of four parts, i.e. basic statistical parameter estimations, basic diagrams drawing, statistical model refereeing, and reliability assessment.Third, software is developed to realize the above framework to complete the treat of reliability test data. Different the existent software of domestic and foreign countries, the new method has been inlaid in the present software.Examples have been given to indicate the availability and the feasibility of the present work.