The Estimation Of Reliability For Satellite Platform With Multiply Type-? Censored Data

The satellite platform is the important part of satellite and supports the performance of payload on satellite.The satellite platform is a series system,which consists of many units.The different satellites consist of the public satellite platform and different payloads.Hence,it is significant to produce satellite platform with high reliability.And the arising problem is the evaluation of the reliability for satellite platform.The lifetime data of satellite platform is mainly from the on-orbit surveillance,which could be transformed to the multiple Type-? censoring model.Besides,the Weibull distribution is widely used to describe the lifetime of products in reliability engineering.Therefore,it is suitable to analyze the reliability of satellite platform using Weibull distribution.There is few research for the inference on the reliability of Weibull distribution with multiply Type-? censored data.In addition,in the existing research,the distribution parameters are usually focused on rather than the reliability.Motivated by these gaps and the demand in engineering,in this paper,the estimations of reliability for satellite platform,including point estimation and lower confidence limit,are considered based on Weibull and exponential distributions with multiply Type-? censored data.In the views of failure data and zero-failure data,the incorporation of lifetime data and other information,products and system,the research is developed as follows:(1)The estimation of reliability for products based on exponential distribution with multiply Type-? censored failure data.For exponential distribution,it is equivalent to provide the estimations of reliability and exponential parameter.First,the maximum likelihood estimate(MLE)of exponential parameter is derived as the point estimation.Then,the lower confidence limits for exponential parameter are constructed using the methods of pivotal quantity,order relations in the sample space,Fisher information and bootstrap.Finally,different limits are compared through the Monte Carlo simulation and application on the computer in satellite platform.The results show that the method of pivotal quantity is superior.(2)The estimation of reliability for products based on Weibull distribution with multiply Type-? censored failure data.First,the MLEs and least-square estimates(LSEs)of Weibull parameters are derived.Accordingly,the point estimations of reliability are obtained using the MLEs and LSEs.For the MLEs,the existence is discussed and the case where the MLEs do not exist is specified.In addition,the approximate maximum likelihood estimates(AMLEs)of Weibull parameters with closed-form are presented.For the LSEs,two methods are provided.Then,the lower confidence limits for reliability are constructed based on the methods of pivotal quantity,Fisher information matrix and bootstrap.The pivotal quantities are given using the LSEs.The limit for the reliability using the method of Fisher information matrix is built through the variance of the estimate of reliability,which is obtained by the covariance of MLEs of Weibull parameters.Finally,through the Monte Carlo simulation and application on the battery of satellite platform,different point estimations and lower confidence limits are compared and analyzed.(3)The estimation of reliability for products with multiply Type-? censored zero-failure data.The reliability is estimated based on Weibull and exponential distributions.Through the match distribution curve(MDC)method,two LSEs of distribution parameters are derived.Accordingly,two point estimations of reliability are obtained.Then,the lower confidence limits for the reliability at the times in the sample are presented and the curve of lower confidence limit is fitted.Because the point estimation and lower confidence limits for reliability are both proposed using the MDC,it is consistent in the application.Finally,through the Monte Carlo simulation together with the applications on gyroscope and GPS receiver,different point estimations and limits are compared to choose the better one.(4)The estimation of reliability for products based on the incorporation of multiply Type-? censored data and other reliability information.The reliability is estimated based on Weibull and exponential distributions through information fusion.First,the prior distribution of distribution parameters is determined based on other reliability information.Next,using the Bayes theory,the posterior distribution of distribution parameters are derived and it is transformed to the one of reliability to obtain the point estimation and lower credible limit for the reliability.Finally,through the Monte Carlo simulation and applications on the products in satellite platform,the point estimation and limit after incorporation are proved to be improved.(5)The estimation of reliability for satellite platform system based on Bayes theory.First,the estimation of reliability for cold-standby system using Bayes theory is studied.Two new methods are proposed based on two existing methods.The posterior distribution of reliability for cold-standby system is derived to obtain the Bayes estimations.Through Monte Carlo simulation and application on the cold-standby system of gyroscope,different methods are compared and the superior one is recommended.Next,the estimation of reliability for series,parallel and k-out-of-n:G systems are discussed.Finally,based on Bayes theory,a method to estimate the reliability of satellite platform system is presented and illustrated through an example.