HE Jian,CAI Jing,XU Kaili,et al.Statistical inference of progressive type II censored under step-stress partial accelerated life test[J].Journal of Yanbian University,2024,(01):13-22，30.
逐步II型截尾步进应力部分加速寿命试验的统计推断
- Title:
- Statistical inference of progressive type II censored under step-stress partial accelerated life test
- 文章编号:
- 1004-4353（2024）01-0013-10
- 关键词:
- 逐步II型截尾样本; 逆Rayleigh分布; 步进应力部分加速寿命试验; 极大似然估计; 近似极大似然估计; 贝叶斯估计
- Keywords:
- progressively type II censored samples; inverse Rayleigh distribution step; stress partial accelerated life test; maximum likelihood estimation; approximate maximum likelihood estimation; Bayesian estimation
- 分类号:
- O213.2
- 文献标志码:
- A
- 摘要:
- 基于逐步II型截尾样本，研究了逆Rayleigh分布在步进应力部分加速寿命试验下的统计分析问题.首先运用贝叶斯方法、极大似然估计方法和近似极大似然估计方法，得到了未知参数和加速度因子的贝叶斯估计(BE)、极大似然估计(MLE)和近似极大似然估计(AMLE)，并运用极大似然估计的渐近正态性推导出未知参数和加速因子的渐近置信区间.其次，运用贝叶斯理论在平方损失函数和线性指数损失函数下给出了模型参数和加速因子的贝叶斯估计和最大后验密度(HPD)置信区间.最后，利用蒙特卡罗模拟对3种不同估计方法的精度进行了评估.数值模拟表明：贝叶斯估计的效果大体上优于极大似然估计和近似极大似然估计的效果；线性指数损失函数下的贝叶斯估计效果要优于平方损失函数下的贝叶斯估计效果；随着样本量的增加，3种估计方法的均方误差(MSE)和平均相对误差(ARE)均呈减小趋势；在同一组样本和同一种移除方案中，BE的HPD置信区间长度优于MLE的渐近置信区间长度.
- Abstract:
- Based on the progressive type II censored samples，the statistical analysis of the inverse Rayleigh distribution under the step-stress partial accelerated life test is investigated. Firstly，Bayesian estimation (BE)，the maximum likelihood estimation (MLE) and approximate maximum likelihood estimation (AMLE) of unknown parameters and acceleration factors are estiblished by using Bayesian method，maximum likelihood estimation method and approximate maximum likelihood estimation method，and their approximate confidence interval is derived by using the asymptotic normality of maximum likelihood estimation. Secondly，the Bayesian estimation and maximum posterior density (HPD) confidence interval of model parameters and acceleration factors are given by using Bayesian theory under square loss function and linear exponential loss function. Finally，the accuracy of three different estimation methods is evaluated by Monte Carlo simulation. Numerical simulation shows that the effect of Bayesian estimation is generally better than that of MLE and AMLE. The Bayesian estimation effect under the linear exponential loss function is better than that under the square loss function. With the increase of sample size，the mean square error (MSE) and average relative error (ARE) of the three estimation methods demonstrated a decreasing trend. In the same group of samples and removal scheme，the HPD confidence interval length of BE is better than the asymptotic confidence interval length of MLE.
参考文献/References:
[1] 鄢伟安，杨海军，周俊杰.基于自适应逐步Ⅱ型混合截尾试验Burr -Ⅻ分布的统计分析[J].系统工程理论与实践， 2020，40(5)：1339-1349.
[2] PATHAK A，KUMAR M，SINGH S K，et al. Bayesian inference for Maxwell Boltzmann distribution on step-stress partially accelerated life test under progressive type-II censoring with binomial removals[J]. International Journal of System Assurance Engineering and Management，2022，13(4)：1976-2010.
[3] PANDEY A，KAUSHIK A，SINGH S K，et al. Statistical analysis for generalized progressive hybrid censored data from lindley distribution under step-Stress partially accelerated life test model[J]. Austrian Journal of Statistics，2021，50(1)：105-120.
[4] FAWZY A M. Inferences under step-stress partially accelerated life tests for linear exponential distribution based on progressive type-II censored data[J].Journal of the Indian Society for Probability and Statistics，2019，20(1)：65-79.
[5] ELSAGHEER R M，MAHMOUND M A W，NAGATY H. Inferences for weibull-exponential distribution based on progressive type-II censoring under step-stress partially accelerated life test model[J]. Journal of Statistical Theory and Practice，2019，13(1)：1-19.
[6] MM Y，REHAB A，G S N. Parametric inference on partially accelerated life testing for the inverted Kumaraswamy distrib-ution based on type-II progressive censoring data[J].Mathematical Biosciences and Engineering：MBE，2023，20(2)：1674-1694.
[7] MUSTAFA K，ALI S S，AHMADUR R，et al. Parameter estimation in step stress partially accelerated life testing under different types of censored data[J].Computational Intelligence and Neuroscience，2022，2022：3491732.
[8] ALAM I，AHMADH H，AHMED A，et al. Inference on adaptive progressively hybrid censoring schemes under partially
[9] AHMED A I. Parametric and interval estimation under step-stress partially accelerated life tests using adaptive type-II progressive hybrid censoring[J]. Annals of Data Science，2020，10(2)：1-13.
[10] 龙兵，张忠占.Ⅰ型双删失下逆Rayleigh分布的统计分析[J].数学的实践与认识，2019，49(06)：199-207.
[11] 李兰平.平方误差和LINEX损失函数下逆Rayleigh分布参数的经验Bayes估计[J].统计与决策，2013(1)：81-83.
[12] ATHIRAKRISHNAN R B，ABDULSATHAR E I. E-Bayesian and hierarchical bayesian estimation of inverse rayleigh distribution[J].American Journal of Mathematical and Management Sciences，2022，41(1)：70-87.
[13] BALAKRISHNAN N，SANDHU A R. A simple simulational algorithm for generating progressive type-II censored samples[J].The American Statistician，2012，49(2)：229-230.
[14] RAQAB M Z，MADI M T. Bayesian inference for the generalized exponential distribution[J]. Journal of Statistical computation and Simulation，2005，75(10)：841-852.
备注/Memo
投稿日期：2023-12-20
基金项目：国家自然科学基金(11901134)第一作者：何剑(1999— )，男，硕士研究生，研究方向为可靠性统计分析.
通信作者：蔡静(1980— )，女，博士，教授，研究方向为可靠性统计分析.