残差控制图应用中的自相关过程模型研究

随着自动化技术的发展,数据自相关现象在现代制造业中普遍存在.基于自相关模型的残差控制图是解决自相关数据统计监控问题的一类较好方法.现有研究均假设数据呈一阶自相关,仅研究一阶自回归模型与残差控制图结合的过程监控问题.但是,实际生产中观测数据可能服从多阶自相关.基于此,针对数据多阶自相关的过程监控问题,运用蒙特卡洛仿真法,...     

控制图在检测实验室质量控制中的应用

0前言 在ISO/IEC 17025<检测和校准实验室能力的通用要求>中,技术要求的第9个要素"检测和校准结果质量的保证"(即5.9条)要求实验室在实际可行的情况下采用统计技术对检测结果数据进行技术核查,以监控检测的有效性,预防问题的发生.目前,各检测实验室正开始积极探索统计技术在检测质量控制方面的应用.本文通过运用 (-X)-R控制图分析钢勾心抗弯刚度的检测数据,说明控制图在检测实验室质量控制中的应用.     

控制图在实验室检测中的应用

本文将统计质量控制方法应用于实验室检测过程中,通过建立检测过程控制图,对检测过程进行评估、监控,判断检测过程是否存在异常,使检测结果得到有效控制。     

具有可变抽样区间的残差MEWMA控制图

为了提高控制图监控多元自相关过程小波动的效率,研究具有可变抽样区间(VSI)的残差MEWMA控制图。首先,对残差MEWMA控制图进行可变抽样区间设计,用蒙特卡罗模拟方法计算其平均报警时间。其次,对所设计的可变抽样区间残差MEWMA与固定抽样区间的残差MEWMA进行比较研究。结果表明,可变抽样区间残差MEWMA控制图监控效率优于固定抽样区间的残差MEWMA控制图,能更快地发现过程中的变化。然后,通过对比研究得出,VSI残差MEWMA控制图的受控性能优于VSI MEWMA控制图。最后,研究平滑系数对过程监控性能的影响,得出过程波动较小时,采用较小的平滑系数;分析自相关系数对过程监控性能的影响关系,得出对于相同波动,相关系数越小,VSI MEWMA残差控制图的监控效果越好。     

控制图技术在卷烟烟气检测分析控制中的应用

作为卷烟产品质量客观评价之一的卷烟主流烟气指标,其检测结果在烟气分析过程中受到温湿度、大气压力和气流速度等测试环境条件;抽吸容量,抽吸时间与间隔等测试仪器条件;以及检测人员等诸多因素影响.有效控制烟气检测结果准确性的影响因素成为必须重视的问题.质量控制过程本身是不稳定的,经常会出现变异.引起这些变异的原因分为随机因素和...     

监控不合格品率的MEWMA-p控制图设计

传统Shewhart-p控制图只对单一属性的不合格品率进行监控,在过程发生偏移时有一定的滞后性。为提高不合格品率控制图的精度,提出一种多元指数加权移动平均不合格品率(multivariate exponentially weighted moving average p, MEWMA-p)控制图。该控制图将多个属性的不合格品率应用于多元指数加权移动平均控制图,可同时对多个属性进行监控,并且对于小范围的偏移更加敏感。对比分析同等偏移程度下指数加权移动平均不合格品率(exponentially weighted moving average p, EWMA-p)控制图与MEWMA-p控制图的平均运行长度(average run length,ARL)结果,并通过模拟仿真说明该方法的有效性。     

基于APL的偏态控制图优化设计

以平均产品长度(APL)为评价控制图性能的标准,研究了偏态控制图的优化设计问题.针对一般控制图无法有效解决偏态总体的不对称性的情况,采用赋权方差法来构造非对称的偏态控制图,并获得其最优设计模型;最后给出了模型的灵敏度分析及算例.     

基于多点报警准则的高质量过程缺陷率控制图

为了提高用于高质量过程监控的累积计点值控制图的监控效率,提出了一类采用多点报警准则的高质量过程缺陷率控制图设计,并给出了计算这类改进型控制图的平均运行长度的马尔科夫链方法。为了验证该设计方法的改进效果,分析比较了采用多点报警规则的3种累积计点值控制图（报警规则为连续两点出界、连续三点中有两点出界、连续三点出界）与报警规则不改变的（报警规则为有一个点出界）累积计点值控制图的监控效率。比较结果表明：在受控状态下平均运行长度都为370的情况下,多点报警的累积计点值控制图的失控状态下的平均运行长度分别较报警规则不变的控制图的平均运行长度减小了约44%、40%、63%,控制图发现缺陷率增大的效率明显改善。得出结论：这种多点报警的累积计点值控制图发现缺陷率增大的速度提高40%~60%。     

平稳自相关过程的EWMA控制图

讨论了对平稳自相关过程中出现的较小波动进行监控的一种方法.采用自回归移动平均(ARMA)模型对平稳自相关过程进行适当的拟合,通过计算残差的方法消除过程中的自相关要素,并在此基础上提出对于均值和方差出现的较小波动进行监控的指数加权移动平均(EWMA)控制图的构造.通过与其它几种方法的比较来说明该方法在监控平稳自相关过程时有更好的效率.     

质量控制图在实验室质量管理中的应用

质量控制图是实验室开展质量控制工作的一种重要工具,便于实验室有效识别检测过程中出现的系统性变异,及时采取措施,使检测过程保持稳定。     

Conditional Control Charts for Weibull Quantiles Under Type II Censoring

In this article, we propose control charts for the quantiles of the Weibull distribution, for type II censored data, based on the distribution of a pivotal quantity conditioned on ancillary statistics. These control charts must be considered as alternatives to bootstrap type control charts. We derive an analytical form of the conditional distribution function of the monitored statistic and we use this function to propose ARL‐unbiased control limits. We further demonstrate that the proposed conditional chart have a general analytical form for the ARL that can be evaluated numerically without use of simulations and we also show that these charts perform at least as well as the bootstrap type ones. We finally apply the conditional charts to a dataset on the strength of carbon fibers to detect shifts in a specified Weibull quantile. Copyright © 2014 John Wiley & Sons, Ltd.     

Monitoring the Weibull Shape Parameter by Control Charts for the Sample Range of Type II Censored Data

In this paper, we propose control charts to monitor the Weibull shape parameter β under type II (failure) censoring. This chart scheme is based on the sample ranges of smallest extreme value distributions derived from Weibull processes. We suggest one‐sided (high‐side or low‐side) and two‐sided charts, which are unbiased with respect to the average run length (ARL). The control limits for all types of charts depend on the sample size, the number of failures c under type II censoring, the desired stable‐process ARL, and the stable‐process value of β. This article also considers sample size requirements for phase I in retrospective charts. We investigate the effect of c on the out‐of‐control ARL. We discuss a simple approach to choosing c by cost minimization. The proposed schemes are then applied to data on the breaking strengths of carbon fibers. Copyright © 2011 John Wiley & Sons, Ltd.     

On the Performance of Two‐sided Control Charts for Short Production Runs

In this paper, statistical performance measures for short‐run production are summarized and proposed. Evaluation and computation based on these performance measures were developed for the two‐sided Shewhart, cumulative sum, and exponentially weighted moving average control charts. Applying these measures, we provided numerical illustrations to compare the efficiency of the three charts and the corresponding Q charts in the cases of known and unknown nominal values of process parameters. Copyright © 2011 John Wiley & Sons, Ltd.     

Performance of Tukey's and Individual/Moving Range Control Charts

This paper compares two control charts: Tukey (TCC) and individual/moving range (XmR) control charts. Both are designed to examine single observation per time period, but little is known about which one is more efficient and under what conditions. We simulated data from different distributions and examined the performance of the two control charts on these data. Performance was assessed using the of average run length, extra quadratic loss, median run length, standard deviation run length, performance comparison index, and relative average run length. Overall, TCC was more efficient than XmR, when observations had binomial, Rayleigh, logistic, lognormal, Maxwell, normal, Poisson, Weibull (with α = 10, β = 1), and Student's t (30 and 10 degrees of freedom) distributions. XmR was more efficient when observations had Student's t (with 4 degrees of freedom) and gamma (with α = 4, β = 1) distributions. These results suggest that improvement teams could reach faster conclusions if they use TCC in most common situations. Copyright © 2014 John Wiley & Sons, Ltd.     

Improved Fast Initial Response Features for Exponentially Weighted Moving Average and Cumulative Sum Control Charts

Exponentially weighted moving average (EWMA) and cumulative sum (CUSUM) control charts have found extensive applications in industry. The sensitivity of these quality control schemes can be increased by using fast initial response (FIR) features. In this paper, we introduce some improved FIR features for EWMA and CUSUM control charts and evaluate their performance in terms of average run length. We compare the proposed FIR‐based EWMA and CUSUM control schemes with some existing control schemes, that is, EWMA, FIR‐EWMA, CUSUM, and FIR‐CUSUM. It is noteworthy that the proposed control schemes are uniformly better than the other schemes considered here. An illustrative example is also given to demonstrate the implementation of the proposed control schemes. Copyright © 2013 John Wiley & Sons, Ltd.     

Nonparametric Synthetic Control Charts for Process Variation

In this article, we propose nonparametric synthetic and side‐sensitive synthetic control charts for controlling fraction nonconforming due to increase in the process variation. Synthetic control chart is a combination of sign and conforming run length control charts. We compare performance of the proposed control charts with the Shewhart sign and S² charts. Our performance study shows that the proposed control charts have a higher power of detecting out‐of‐control signal. We also study the steady‐state behavior of a nonparametric synthetic control chart. We present a Markov chain model to evaluate the steady‐state average run length of the synthetic and side‐sensitive synthetic control charts. Copyright © 2011 John Wiley & Sons, Ltd.     

Some Recent Developments on the Effects of Parameter Estimation on Control Charts

Statistical process control plays a key role in today's highly competitive industrial environment since it allows quality practitioners to timely detect out‐of‐control situations and take actions whenever necessary in order to ensure that the products or services produced correspond to certain quality standards. Control charts are the tools quality practitioners use, and their monitoring performance is of major importance in practical applications. Since the values of the parameters used for the design of the charts' control limits are usually unknown in practice, the practitioners need to estimate them using an in‐control retrospective sample. It has been shown that parameter estimation severely affects the control charts' properties. Many recent studies focused on investigating the impact of parameter estimates on the performance of control charts and on ways of diminishing this impact. This paper aims to provide an up‐to‐date critical review on the methodologies that have recently been developed in this area. Copyright © 2013 John Wiley & Sons, Ltd.     

Mixed Exponentially Weighted Moving Average–Cumulative Sum Charts for Process Monitoring

The control chart is a very popular tool of statistical process control. It is used to determine the existence of special cause variation to remove it so that the process may be brought in statistical control. Shewhart‐type control charts are sensitive for large disturbances in the process, whereas cumulative sum (CUSUM)–type and exponentially weighted moving average (EWMA)–type control charts are intended to spot small and moderate disturbances. In this article, we proposed a mixed EWMA–CUSUM control chart for detecting a shift in the process mean and evaluated its average run lengths. Comparisons of the proposed control chart were made with some representative control charts including the classical CUSUM, classical EWMA, fast initial response CUSUM, fast initial response EWMA, adaptive CUSUM with EWMA‐based shift estimator, weighted CUSUM and runs rules–based CUSUM and EWMA. The comparisons revealed that mixing the two charts makes the proposed scheme even more sensitive to the small shifts in the process mean than the other schemes designed for detecting small shifts. Copyright © 2012 John Wiley & Sons, Ltd.     

A General Framework for Constructing Control Charts

A general framework for the construction of control charts is presented. The method is based on using the density of the sample subgroup statistic as a measure of how unusual newly observed subgroups are. This methodology includes, as special cases, many common control chart techniques. The method is also easily applied to multivariate and multimodal situations. A non‐parametric control chart is implemented by estimating the density of the sample subgroup statistic using a kernel estimator of the bootstrap distribution of the observed subgroup statistics. Several examples of the method are presented in the parametric and non‐parametric situations. The potential performance of the non‐parametric version of the method is demonstrated through an empirical study of average run length properties. Copyright © 2005 John Wiley & Sons, Ltd.