Research On Methods Of Reliability Assessment With Small Sample Test On The Basis Of Bayes Theory

The reliability analysis and evaluation for complex system such as intercontinental ballistic missile and satellite is difficult in engineering because of small sample tests. Bayesian method can fuse various prior information effectively, and assess the reliability of weapon system combining a little field information. Therefore, according to the actual reliability characteristic of weapon system, on the basis of Bayes theory, the evaluation method for reliability with small sample is studied in this paper .The central work of this paper are as follows:First, this paper discusses the prior information pretreatment, including how to collect and convert prior information, methods of distribution type test and compatibility test of prior information, and so on. Then the reliability bayes assessment methods for life distribution types in common use also are discussed,such as exponential life model, Weibull life model, normal life model and log-normal life model.And then, for the traditional methods cannot differentiate the prior information and the field information, this paper investigates fusion of information form multiple sources in bayesian assessment for reliability of success or failure product. This model uses mixed prior distribution, and determines the inheritance factor by using chi-square goodness of fit. Its rationality is easy to be comprehended.Whereafter, for many complex systems lacked failure information, this paper improves the performance reliability synthesis method. It introduces Bootstrap method into the system performance reliability statistical analysis, Bootstrap method is data-based simulation to carry out familiar statistical calculations ,so it is practicality for it is purely computation means rather than using of statistical formulas. On the basis of all the foregoing research, the assessment method of system reliability using the performance information is investigated in this paper.Finally, avaliability and practicality of the methods studied in this paper were indicated by the analysis to an example.