高删失电机寿命数据的贝叶斯可靠性分析  

Bayesian reliability analysis of high-censoring motor lifetime data

作　　者：潘竞翔 袁晓惠 张欣然 PAN Jingxiang;YUAN Xiaohui;ZHANG Xinran(School of Mathematics&Statistics,Changchun University of Technology,Changchun 130012,China)

机构地区：[1]长春工业大学数学与统计学院,吉林长春130012

出　　处：《长春工业大学学报》2024年第4期345-352,共8页Journal of Changchun University of Technology

基　　金：吉林省科技厅重大科技专项(20210301038GX);吉林省教育厅科学研究项目一般项目(JJKH20230749KJ)。

摘　　要：为了提高具有高删失特征寿命数据的评估精度,得到失效概率的区间估计以及分布模型的点估计,文中提出基于MCMC的Gibbs算法结合舍选法来模拟失效概率的估计方法,并计算其置信区间。首先考虑实际情况中失效概率的次序特征,在贝叶斯的理论下通过失效概率的取值范围和舍选法构建超参数优化模型;然后基于似然方程抽取失效概率。在仿真算例与具有高删失特征的电机结构寿命数据中,证明了该方法在威布尔分布模型的假设下,对失效概率的评估效果在精度方面优于E-Bayes方法,能更好地拟合数据。In order to enhance the assessment accuracy of high-censoring feature lifetime data and obtain interval estimates of failure probability as well as point estimates of distribution models,this paper proposes an estimation method for simulating failure probabilities based on the MCMC Gibbs algorithm combined with the rejection method.It also calculates confidence intervals.Firstly,the ordinal characteristics of failure probabilities in practical scenarios are considered.Under the framework of Bayesian theory,a hyperparameter-optimized model is constructed by taking into account the range of failure probabilities and employing the rejection method.This approach helps capture the features of failure probabilities more effectively.Subsequently,failure probabilities are extracted based on the likelihood equation.Through simulations and the analysis of lifetime data from motor structures with high-censoring features,it has been demonstrated that this method,under the assumption of the Weibull distribution model,outperforms the E-Bayes method in terms of the accuracy of failure probability estimation.It provides a better fit to the data.

关 键 词：高删失特征 电机寿命 可靠性 贝叶斯分析 舍选Gibbs方法 

分 类 号：O212[理学—概率论与数理统计]