基于贝叶斯推断的威布尔型产品可靠性评估  

Reliability Assessment of Weibull Products Based on Bayesian Inference

作　　者：何成铭 边云龙 HE Cheng-ming;BIAN Yun-long(Department of Equipment Support and Remanufacturing,Army Academy of Armored Forces,Beijing 100071,China)

机构地区：[1]陆军装甲兵学院装备保障与再制造系,北京100071

出　　处：《价值工程》2022年第27期149-152,共4页Value Engineering

摘　　要：随着制造工艺与管理水平的提高,产品可靠性水平同样得到很大提升,随之带来故障数据短缺的问题又限制了对高可靠性产品的评估,可靠性评估面临小子样问题。针对上述问题应用贝叶斯推断建立威布尔型产品可靠性模型,充分利用各种先验信息与专家经验生成先验概率密度,结合样本数据所建立的似然函数生成后验概率密度,用马尔可夫链蒙特卡洛法求解待估参数,既扩大了贝叶斯推断的应用范围,也保证了求解精度。最后以某风电机组齿轮箱故障数据为算例,说明了本文提出模型的有效性与准确性。With the improvement of manufacturing process and management level,the level of product reliability has also been greatly improved.The problem of shortage of fault data has also limited the assessment of high-reliability products.Reliability evaluation is faced with a small problem.Aiming at the above problems,Bayesian inference is used to establish a Weibull-type product reliability model,and various prior information and expert experience are fully utilized to generate prior probability density,and the likelihood function established by sample data is combined to generate posterior probability density.The Markov Chain Monte Carlo method solves the parameters to be estimated,which not only expands the application scope of Bayesian inference,but also ensures the accuracy of the solution.Finally,the effectiveness and accuracy of the model proposed in this paper are illustrated by taking the fault data of a wind turbine gearbox as an example.

关 键 词：可靠性评估 贝叶斯推断 马尔可夫链蒙特卡洛法 无掉头抽样器(NUTS) 

分 类 号：O213.2[理学—概率论与数理统计] N945.17[理学—数学]