Reliability Estimates And Lower Confidence Limits Of Several Systems

Reliability is the probability of the product to perform a required function in a predetermined period of time and a predetermined condition. Reliability of the system can provide methods and theoretical basis for the improvement the reliability of the product, analyze the reasons of the failure of the product, identify weaknesses, suggest improvements, strengthen quality management and assess whether the level of reliability can meet the design requirements. Further more, it can also provide management information for production and decisions, provide help for people’s daily life, as well as methods and means for product innovation. Confidence limits, especially the study of confidence lower limit as a very important issue in system reliability studies, therefore, is not only very necessary but of great significance both in theory and application.This paper is mainly composed of three parts:Part One:In the case of zero-failure data, calculate out the new estimation formula about the reliability of double-parameter distribution by using a new modified maximum likelihood estimation method and Bayesian estimation method. Finally calculation is performed concerning practical problem. Part Two:Based on the Fiducial theory, the expressions of the reliability in the parallel-series system which composed of Poisson components and pass-fail switches were obtained, and based on fixed number testing data of components and pass-fail testing data of switches, the computer simulation by means of Beta and Monte-Carlo was obtained,confirmed that the combination of the Fiducial theory and the Beta fitting method is feasible and practical.Part Three:Based on the Fiducial theory, the expressions of the reliability in the series-parallel system which composed of Poisson components and pass-fail switches were obtained, and based on fixed number testing data of components and pass-fail testing data of switches, the computer simulation by means of Beta and Monte-Carlo was obtained confirmed that the combination of the Fiducial theory and the Beta fitting method is feasible and practical.