Statistical Inference Of Generalized Half-normal Distributions Based On Constant Accelerated Life Tests Of Geometric Processes

In this thesis,we study statistical inference on the parameters of generalized halfnormal distribution under the constant accelerated life test.Firstly,by using the random monotonicity of geometric process,we introduce different stresses to obtain a new model of constant accelerated life test.Secondly,we study statistical inferences on the parameters of generalized half-normal distribution and the ratio of geometric process under the complete sample,the type I censored and type II censored sample,respectively.The approximate values of maximum likelihood estimations of parameters are obtained by Newton iteration method.By asymptotic normality of maximum likelihood estimators,we obtain asymptotic confidence interval estimations of parameters.The bootstrap confidence interval estimations of the parameters are given also.The ratio of geometric process is tested by hypothesis tests.Finally,we program and numerical simulation by R software.The simulation results show that the parameter estimations are effective and the model hypothesis are rational.