The Research On Prior Information Credita-bility Of Bayes Reliability Estimation

High-tech weapon equipments have the characteristics of small-sample filed testand multi-source prior information, and it is difficult for them to do the reiliability test.Bayes statistics verification test method can make full use of multi-source informationto do the equipment reliability estimate in small-sample situation. However, when theinformation used in Bayes reliability estimate has a low value of creditability, the resultsof reliability estimate may be reduced, what’s worse, the vast fuzzy prior informationmay “flood” the field test information, and lead to the inaccurate Bayes reliability esti-mate results. Therefore, in order to enhance the veracity of Bayes reliability estimateresults, this article did a further research on the creditability of prior information.Firstly, this paper did a research on the definition and calculation of the creditabilityof prior information and proposed a Prior Information Creditability Calculation Modelbased on Outlier Test (CMOT). Aiming at the Bayes reliability estimate type of priordistribution and field test data, using prior distribution simulates multigroup populationdistributions, making use of these distributions to do outlier test on field test data, andcounts the number of none outlier test. Finally, calculate the prior information credita-bility as the ratio of none outlier test number and whole test number.Secondly, the article did an analysis of rationality judgment of prior informationcreditability and gave a Prior Information Creditability Rationality Judgment Modelbased on Simulation (CJMS). Aiming at results uncertainty of the prior informationcreditability, we need to propose corresponding rational decision rules based on certainconfidence level. It is necessary to make sure the prior information satisfies given ra-tional decision rules, if not, the prior information should not be used into the Bayes re-liability statistical test. Based on different confidence level, the probability densityfunction curve gave the low limit of prior information creditability.Thirdly, this paper analyzed the influence of prior information creditability and-structured a Prior Information Impact Model based on Location Parameter Translation,(IMLT). Based on the influence relationship between location parameter and prior in-formation creditability, this model confirms the excursion type of location parameterand adjusts the location parameter to revises the prior information. When the pripr in-formation creditability reaches the maximum, the corresponding prior information isthe most accurate information. Using the above prior information to do the Bayes reli-ability estimation and analysis can fully enhance the accuracy of Bayes statistical re-sults.Finally, we make a conclusion of the work and explicits the further research.