基于部分线性模型的profile监控-统计学专业论文.docxVIP

摘要产品的某种质量特征可以由一个响应变量和一个或多个解释变量之间的函数关系来刻 摘要 产品的某种质量特征可以由一个响应变量和一个或多个解释变量之间的函数关系来刻 画，这类函数关系常常用profile来描述，对此类profile的监控是当今SPC领域的热点研究方 向之一。本文基于多元指数加权移动平均方法，结合半参数模型的优势，提出了一种当过 程可以用部分线性模型来描述时的统计过程控制方法，该方法不仅有效地解决了部分线性 模型下profile自4]监控问题，而且相较于非参数模型下profile的监控，一定程度上提高了效率， 同时不再受参数方法中误定失控模型带来的困扰。通过大量数值模拟，比较了所提出的方 法与现有方法监控效率的优劣。此外，本文还给出了一种系统的统计诊断方法，来判断变 点的位置，识别模型的具体变化。最后通过对一个电力负荷实际数据的分析，进一步验证 了所提出方法的有效性。 关键词：指数加权移动平均广义似然比检验部分线性模型profile监控统计过程控制 万方数据 AbstractQuality Abstract Quality of process is characterized by the functional relationship between response variable and more explanatory variablesj such functional relationship is usually described by profile monitoring such profiles is of the hot research direction in the field of SPC．According the advantage of semi—parametric models，we propose novel scheme to monitor partial linear profiles based multivariate exponentially weighted moving average procedure．The proposed scheme not only provides more effective SPC solution to solve partial linear profiles than nonparametric profiles：but also resolves the latent problem in popular parametric monitoring methods of being unable to detect certain types if changes due to misspecified，out—of-control model．Our simulation results compare method with former in AREs．In addition，a systematic diagnostic approach is provided to locate the change point of the process and identify the type of change in the profile． Finally,a power load example is used to illustrate the implementation of the proposed monitoring and diagnostic approach Key words：Exponentially weighted moving average，Generalized likelihood ratio test，Partial linear models，Profile monitoringj Statistical process control 万方数据 目 目 录 摘 要 ； Abstract 第一章绪 论 1 1．1问题背景．．．．．．．．，．．．．．．．．．．．．．．．．．．．．．．．．．．．．．．．．． ● 1．2部分线性模型的研究现状．．．．．．．．．．．．．．．．．．．．．．．．．．．．．．．． 。 1．3 profile监控的研究进展及其应用．．．．．．．．．．．．．．．．．．．．．．．．．．．． 2 1．4本文的主要工作．。。，．．．。。．．，。．．．．．。。。．．。．．。．。．。．。．。．。． 3 第二章统计过程控制的预备知识 5 2．1统计过程控制的概述．．．．．．，。．．．．．．。．．．．．．．．．．．。。．．．．，，。 5 2．2各种控制图的