基于似然比检验的多项式控制图

发布时间：2019-01-05 19:21

【摘要】：随着工业领域自动化的高度敏度传感器等设备的普及应用,信息服务业领域的高性能计算机技术的发展.在线大规模、高维、复杂数据快速收集得以实现.在许多问题中,我们所关心的产品质量已经不是简单的某个或某几个指标的均值或方差(或分布),而是需要用一些自变量和响应变量之间的关系才能更好的刻画或衡量,这种关系通常称为曲线(profile)关系.很多情况下,这种曲线关系可以通过多项式的回归关系来描述.本文中,我们提出了一种基于似然比检验的第II阶段(Phase II)多项式控制图,通过平均运行长度(ARL)来衡量控制图的性能.并与正交多项式(OTH)控制图、多元累计加和(MCUSUM)控制图、多元指数加权移动平均(MEWMA)控制图、T~2控制图进行比较.模拟结果表明,本文提出的多项式控制图具有很好的检测能力.
[Abstract]:With the popularization and application of automatic high sensitivity sensors and other devices in the industrial field, the development of high performance computer technology in the field of information service industry. Online large-scale, high-dimensional, complex data collection can be achieved. In many problems, the product quality we are concerned with is not simply the mean value or variance (or distribution) of one or more indicators, but the relationship between some independent variables and response variables to be better characterized or measured. This relationship is usually referred to as the curve (profile) relation. In many cases, this curve relationship can be described by polynomial regression relations. In this paper, we propose a II stage (Phase II) polynomial control chart based on likelihood ratio test. The average running length (ARL) is used to evaluate the performance of the control chart. It is compared with orthogonal polynomial (OTH) control chart, multivariate cumulative addition and (MCUSUM) control chart, multivariate exponentially weighted moving average (MEWMA) control chart and Tap2 control chart. The simulation results show that the polynomial control chart presented in this paper has good detection ability.
【学位授予单位】：辽宁大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O213.1

【相似文献】