基于自相关过程的贝叶斯质量控制模型研究  

Bayesian quality control model based on autocorrelation stochastic process

作　　者：朱慧明[1] 赵锐[1] 

机构地区：[1]湖南大学工商管理学院,湖南长沙410082

出　　处：《计算机集成制造系统》2008年第3期615-618,624,共5页Computer Integrated Manufacturing Systems

基　　金：教育部新世纪优秀人才支持计划资助项目(NCET050704);湖南省自然科学基金资助项目(05JJ30130)~~

摘　　要：为解决工序质量控制中自相关过程的观测值并不满足控制变量独立性的基本假设问题,提出了贝叶斯质量控制方法。应用基于Gibbs抽样的马尔可夫链蒙特卡罗方法模拟模型参数的后验分布,构建了自相关过程的贝叶斯统计质量控制模型,使得拟合后的残差序列具有相互独立性质,从而满足常规控制图的基本假设前提,避免了在受控状态下使用常规控制图造成的漏发或虚发报警现象。研究结果表明,贝叶斯方法是解决自相关条件下质量控制的有效工具。To solve the problem that the data generated by autocorrelative process couldn＇t meet the basic assumption of variables＇ independence in quality control, Bayesian quality control method was introduced. And the posterior distributions of parameters in quality model were simulated via Markov chain Monte Carlo based on Gibbs sampling, by which the Bayesian statistical quality control model was set up with the autocorrelative data. Residual series in the model were independent so as to satisfy the necessary assumptions in quality control charts. The Bayesian model could avoid incorrect alarming when the process was under control, or not alarming when the process was out of control. Results showed that the Bayesian method was an effective tool to monitor the quality control of autocorrelation process.

关 键 词：质量控制 贝叶斯估计 蒙特卡罗方法 马尔可夫过程 仿真 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]