Statistical Processing Methods And Software Development Of Life Test’s Censored Data

It is very inconceivable if without life test about reliability studies, but it’s not allowed if the time about life test is too long. This not only leads to increased testing costs, but also let the test expire let to lose its significance. So, it is a very worthy study that shortens the life test of time. So, this involves censoring of life. This is the question to be studied in this thesis how to estimate the probability of failure and lifetime of the specimen according to censored data. In other words, the main research in this thesis is statistical methods of censoring life test data.This paper proposes three distribution models about censoring life test data, and also carried out a detailed analysis and discussion separately for each distribution model.For the exponential distribution model, this paper respectively studied type i censoring data and type ii censoring data, no matter replacement data and no replacement data still single-parameter and two-parameter. This paper respectively given the maximum likelihood estimation of θ, reliability estimation and the estimation of failure rate λ. For the Weibull distribution model, this paper respectively studied Two-parameter Weibull distribution model and Three-parameter Weibull distribution model, and also given their parameter estimation. Studies on traditional methods, for example, probability map estimation method, also studied the improvement of existing methods, for example, inverse moment estimation method etc. For the normal and lognormal, this paper discussed the parameters estimate under the random censoring with lognormal; also it’s used maximum likelihood estimation and iteration, etc.Owing to it is very abnormal cumbersome of the mathematical and statistical process about test data, so this paper written corresponding software, in respect of the above three distribution models were collate, facilitate future using and secondary development.