The Research On Bayesian Reliability Evaluation Based On Maximum Entropy And Markov Chain Monte Carlo Method

21 Century is the era of quality,as the core and foundation of quality universal characteristics,the importance of reliability speaks for itself.The implementation of "Six Sigma",the concept of "no maintenance period(MFOP)" and"reliability 3.0:put forward new opportunities and challenges for the development of reliability.Product development time is shorter,the actual market is more urgent,all of these have put forward new requirements for the development of reliability.In order to investigate the reliability evaluation of small samples,this paper systematically studies the application of maximum entropy method(MEM)and Markov Chain Monte Carlo(MCMC)method in Bayesian Reliability Evaluation framework,to solve the problem that the traditional statistical method fails to evaluate the reliability under the small sample failure characteristics.Aiming at small sample failure characteristics,reliability evaluation under bayesian framework is divided into prior information collection and inspection based on prior conversion,the maximum entropy method to determine the distribution,based on Markov chain Monte Carlo method of posterior distribution calculation and reliability evaluation based on the posterior distribution of four parts.In this paper,the reliability evaluation of small sample problem,each stage will be the maximum entropy method and MCMC method are respectively applied to the reliability of small sample evaluation under the Bayesian framework,maximum entropy and multi-source information fusion,improve the utilization efficiency of the prior information to reduce subjective factors in the process of determining prior distribution.MCMC method for solving the reliability of the posterior distribution is a good solution to the posterior distribution of complex and high dimensional numerical integration to solve the problem,the Bayesian method has a wider application space in the small sample reliability assessment.