The Bayesian Reliability Assessment and Prediction for Radar System Based on New Dirichlet Prior Distribution

8th International Conference on Advanced Infocomm Technology

Ming, Zhimao^1 ; Ling, Xiaodong^1 ; Bai, Xiaoshu^1 ; Zong, Bo^1

China Satellite Maritime Tracking and Control Department, Jiangsu

214431, China^1

关键词: Bayesian reliabilities;    Bayesian reliability assessment;    Constraint conditions;    Dirichlet distributions;    Dirichlet prior distribution;    Multidimensional numerical integration;    Posterior distributions;    Uniform distribution;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/679/1/012039/pdf
DOI  :  10.1088/1742-6596/679/1/012039

来源: IOP

【 摘 要 】

This article studies on Bayesian reliability growth models of complex system based on new Dirichlet prior distribution when the sample of system is small. The model briefly describes expert experience as uniform distribution, then equivalent general Beta distribution of uniform distribution can be solved by optimization method when prior parameters are variables, mean is constraint condition, and variance is regarding as the optimization objective. The optimization method solves the problem of how to determine values of hyper-parameters of new Dirichlet distribution when these parameters have no specific physical meaning. Because the multidimensional numerical integration of posterior distribution is very difficult to calculate, WinBUGS software is employed to establish Bayesian reliability growth model based on a new Dirichlet prior distribution, and two practical cases are studied under this model in order to prove validity of model. The analysis results show that the model can improve the precision of calculation, and it is easy to use in engineering.