Modeling And Research On Reliability Analysis Of Key Components Of High Speed Emu Braking System For Small Sample Problem

Reliability assessment of lifetime of the product is significant for enhancing the life,making the maintaince project, reducing the cost and improving the market competitiveness.The reliability of the key components of high speed EMU braking system directly affects the braking performance of vehicle. And there is a close relationship between the accuracy of reliability assessment and safety of vehicle during service. From the traditional method, the reliability assessment is carring on the statistics to the life of a large number of samples(usually n>30). This means that it is necessary for the reliability test to spend a lot of time and money. However, It has a so long design service life, expensive cost, complex structure for key components of braking system of high speed EMU that it is impossible for reliability assessment using a traditional method based on big data theory. In order to solve this problem,a method for analyzing the reliability under small sample circumstances is required. In this paper, the reliability of key components of high speed EMU braking system as the research object are studied under small sample and censored data using Bayes method.Generally speaking, this paper mainly discusses the contents as follows:1. Some of the failure types and characteristics of the key compoents of high speed EMU braking system are listed and analyzed. And then a comparsion between Modified Weibull distribution adn common lifetime distribution is discussed for the key components. After that,It is explained for appling Bayes theory to solve the problem of reliability assessment under samll sample.2. Aiming at the reliability assessment for key components of high speed EMU braking system under small sample conditions with mixed censoring lifetime test and zero failure data,the two kinds of Bayes method which are "The reliability model based on Modified Weibull distribution" and "The reliability model based on Binomial distribution assumption Modified Weibull lifetime " are presented and researched in details, respectively. And also the principle of choicing prior distribution, the hierachical Bayesian mehthod, the approach for defining hyper parameters, the relationship among failure probability for each time point based on the nature of convex function, the derivation process of posterior distribution for two model and the robustness of model are discussed.3. For two Bayes model above, the hybrid MCMC algorithm is designed combining Gibbs sampling and Metropolis-Hasting sampling in improving the sampling efficiency.4. Using the above two models, the reliability of relay valve and Emergency brake solenoid valve of high speed EMU braking system is researched and the relavant estimators of posterior distribution are obtained. Then, the model selection and robustness of model are studied. Through the two case, the rationailty for the above two kinds of model for reliability are demonstrated and high efficiency for hybrid MCMC algorithm is proved. The research work can provide the theoretical basis for reliability assessment for key components of high speed EMU braking system under small sample conditions.