| We summarize the application of the changing shape and scale model(Chss) proposed by Bagdonavicius and Nikulin(1999)[1] in the accelerated life testing(ALT) process,CHSS model is an Expandation of the accelerated failure time model[2],The CHSS model allow the reliable experiment not only of the parametic but also the non-parametic estimating process. Chapter One is the introduction, which introduce the structure of the entire article and literature review. In the second chapter introduces the CHSS model of background knowledge and the selection of experimental program.Firstly, previous work made on the ALT is given, the main points are to consider three cases:(a) Survival function S x0 and transfer function r are parameteried;(b) Unknown survival function S x0 and transfer function r are parameteried;(c) Unknown survival function S x0 and transfer function r .Then given a broader application in the ALT accelerated failure model: and step-stress model and for the AFT model in practice prone to error and wide scope of application defects without CHSS model propose improved method. In this article, decide to take the CHSS model which is natural expansion of AFT model, the experimental program: in the accelerated pressure take the units with very small coefficient of variation on individual tests and combined with the test results under normal pressure to amend.Chapters Three and Four are the main body of the paper, which mainly estimate the parameters and non-parameter of the CHSS model and give the reliability of parameters of CHSS model under multiple pressures.By the premise of the experimental plan proposed in§2.2,we reformulate the CHSS (Changing Shape and Scale),Bagdonavicius and Nikulin,(2002)[1] at step-stress condition: . Suppose the survival function at normal pressure only decided by the shape and the shape parameter: is the survival function with free parameters.So we consider CHSS model like For mentioned experimental method the survival function with stress x1 and Denote (?) the maximum likelihood estimation of ?, then we estimate the survival function at nominal stressIn the non-parametric case, the modified maximum likeli- hood estimators take the minimum of (?) with the parametersαand r.Denote (?) the modified maximum likeli- hood estimate ofα,r,then the estimate of survival function at nominal stressAt multiple stress estimate of CHSS model with Weibull distribution,exchangeλandηwith (?) and (?) and obtain the formulate of the reliability estimate of R(Chapter Four):In the end, I summarise references to the CHSS model proposed estimation method and analyze the experimental data on the estimation methods. |
