Fast Initial Response for CUSUM Quality-Control Schemes: Give Your CUSUM A Head Start

The fast initial response (FIR) feature for cumulative sum (CUSUM) quality-control schemes permits a more rapid response to an initial out-of-control situation than does a standard CUSUM quality-control scheme. This feature is especially valuable at start-up or after a CUSUM has given an out-of-control signal. This article presents the average run length and the distribution of run length for CUSUM schemes with the FIR feature and compares FIR CUSUM schemes to standard CUSUM schemes. The comparisons show that if the process starts out in control, the fast initial response feature has little effect; however, if the process mean is not at the desired level, an out-of-control signal will be given faster when the FIR feature is used.     

A Neural Network Method for Modelling the Parameters of a CUSUM Chart

The cumulative sum (CUSUM) chart is widely employed in quality control to monitor a process or to evaluate historic data. CUSUM charts are designed to exhibit acceptable average run lengths both when the process is in and out of control. This paper introduces a functional technique for generating the parameters h and k for such a chart that will have specified average run lengths. It employs the method of artificial neural networks to derive the appropriate coefficients. An EXCEL spreadsheet to assist computing the parameters is presented.     

An Evaluation of the Crosier's CUSUM Control Chart with Estimated Parameters

We evaluate the performance of the Crosier's cumulative sum (C‐CUSUM) control chart when the probability distribution parameters of the underlying quality characteristic are estimated from Phase I data. Because the average run length (ARL) under estimated parameters is a random variable, we study the estimation effect on the chart performance in terms of the expected value of the average run length (AARL) and the standard deviation of the average run length (SDARL). Previous evaluations of this control chart were conducted while assuming known process parameters. Using the Markov chain and simulation approaches, we evaluate the in‐control performance of the chart and provide some quantiles for its in‐control ARL distribution under estimated parameters. We also compare the performance of the C‐CUSUM chart to that of the ordinary CUSUM (O‐CUSUM) chart when the process parameters are unknown. Our results show that large number of Phase I samples are required to achieve a quite reasonable performance. Additionally, the performance of the C‐CUSUM chart is found to be superior to that of the O‐CUSUM chart. Finally, we recommend the use of a recently proposed bootstrap procedure in designing the C‐CUSUM chart to guarantee, at a certain probability, that the in‐control ARL will be of at least the desired value using the available amount of Phase I data. Copyright © 2015 John Wiley & Sons, Ltd.     

On the effectiveness of the unit sample size for standard control charts: ARL VS. Aurl

In this paper we investigate the use of the average unit run length (AURL) as an important measure of the effectiveness of various quality control charting schemes. In particular we focus on its appropriateness for normally distributed processes that tend to produce units (or measurements) at slow rates. In our investigations with the standard Shewhart X? and R charts, as well as the CUSUM chart, AURL shows that a sample size of n=1 can yield the fastest means of detecting shifts.     

Using CUSUM Control Schemes for Monitoring Quality Levels in Compound Poisson Production Environment: The Geometric Poisson Process

In contemporary modern and high volume production environments such as wafer manufacturing, a small sustained shift is not very easily detected in a short period of time, but may have a great impact on a manufacturing process. Thus, it is important to be able to detect and identify a small sustained shift of the production process in a timely manner and correct the undesired situation. The cumulative sum (CUSUM) control scheme is considered to be one of the efficient reference tools in detecting a small structure change in a process. However, for control of defects in a production process, too often the assumption is made that the defects follow a Poisson distribution. In practice, the process is more complex and the distributions of defects are more appropriately modeled by the compound Poisson distribution. In this paper, the underlying distribution is the geometric Poisson distribution, a Poisson distribution compounded by a geometric distribution, and the CUSUM control scheme based on the geometric Poisson process is addressed. An effective CUSUM control scheme can provide an adequate average run length (ARL), that can be obtained from the probability transition matrix for the Markov chain proposed by Brook and Evans (1972). With proper ARL selected, the geometric Poisson CUSUM control scheme is developed for process control.     

New CUSUM and dual CUSUM mean charts

The CUSUM (C) charts are well recognized as a potentially advanced process monitoring tools because of their sensitivity against small and moderate shifts. In this paper, we first improve the sensitivity of the Brownian motion–based C (BC) chart with an appropriate transformation, named new BC (NBC) chart, for monitoring moderate and large shifts in the mean of a normal process. Then, using the control charting structure of the Crosier C (CC) chart, we propose the NBCC (NBC with CC structure) chart. In addition, for efficiently detecting a mean shift within an interval, dual version of these control charts are also proposed, named the dual NBC (DNBC) and dual NBCC (DNBCC) charts. Moreover, the fast initial response feature is also incorporated into the proposed charts. Using the Monte Carlo simulation, the run length properties of the proposed charts are computed. The run length performances of the existing and proposed charts are compared using the extra quadratic loss and integral relative average run length as performance criterion. It turns out that the NBC and NBCC (DNBC and DNBCC) charts are uniformly more sensitive than the C, CC, and NBC (dual C and dual CC) charts when detecting the mean shifts in small, moderate, and large intervals, where the DNBCC chart outperforms all considered charts. The proposed charts are also applied on real data sets to support the proposed theory.     

Progressive Mean Control Chart for Monitoring Process Location Parameter

Control charts are widely used for process monitoring. They show whether the variation is due to common causes or whether some of the variation is due to special causes. To detect large shifts in the process, Shewhart‐type control charts are preferred. Cumulative sum (CUSUM) and exponentially weighted moving average (EWMA) control charts are generally used to detect small and moderate shifts. Shewhart‐type control charts (without additional tests) use only current information to detect special causes, whereas CUSUM and EWMA control charts also use past information. In this article, we proposed a control chart called progressive mean (PM) control chart, in which a PM is used as a plotting statistic. The proposed chart is designed such that it uses not only the current information but also the past information. Therefore, the proposed chart is a natural competitor for the classical CUSUM, the classical EWMA and some recent modifications of these two charts. The conclusion of this article is that the performance of the proposed PM chart is superior to the compared ones for small and moderate shifts, and its performance for large shifts is better (in terms of the average run length). Copyright © 2012 John Wiley & Sons, Ltd.     

Computing the Percentage Points of the Run-Length Distributions of Multivariate CUSUM Control Charts

Multivariate CUSUM control charts are often used instead of the standard Hotelling's control charts in many practical problems when detection of small shifts in the process mean is important. However, design of multivariate CUSUM control charts are usually based on the average run length (ARL). In this work, we will compute the percentage points of the run-length distributions of two multivariate CUSUM control charts. It will be shown that interpretations based on ARL can be misleading since the in-control run-length distribution of a multivariate CUSUM is highly skewed. On the other hand, the percentage points of the run-length distribution provide additional information such as the median run length, early false out-of-control signals, and the skewness of the run-length distribution for a particular scheme. These extra information might provide quality control engineers further knowledge of a particular multivariate CUSUM control chart scheme.     

Improving the performance of CUSUM charts

The control chart is an important statistical technique that is used to monitor the quality of a process. Shewhart control charts are used to detect larger disturbances in the process parameters, whereas CUSUM and EWMA charts are meant for smaller and moderate changes. Runs rules schemes are generally used to enhance the performance of Shewhart control charts. In this study, we propose two runs rules schemes for the CUSUM charts. The performance of these two schemes is compared with the usual CUSUM, the weighted CUSUM, the fast initial response CUSUM and the usual EWMA schemes. The comparisons revealed that the proposed schemes perform better for small and moderate shifts, whereas they reasonably maintain their efficiency for large shifts as well. Copyright © 2010 John Wiley & Sons, Ltd.     

A Group of Moving Averages Control Plan for Signaling Varying Location Shifts

Control plans consisting of a group of moving averages (GMA plans) of various sizes are proposed for monitoring processes. Since moving averages of different sizes retain several levels of memory of past observations, these plans have good average run length (ARL) properties over a range of location shifts. The ARLs of GMA plans are compared with the conventional cumulative sum and the exponentially weighted moving average procedures.     

A distribution-free tabular CUSUM chart for autocorrelated data

A distribution-free tabular CUSUM chart called DFTC is designed to detect shifts in the mean of an autocorrelated process. The chart's Average Run Length (ARL) is approximated by generalizing Siegmund's ARL approximation for the conventional tabular CUSUM chart based on independent and identically distributed normal observations. Control limits for DFTC are computed from the generalized ARL approximation. Also discussed are the choice of reference value and the use of batch means to handle highly correlated processes. The performance of DFTC compared favorably with that of other distribution-free procedures in stationary test processes having various types of autocorrelation functions as well as normal or nonnormal marginals.     

Robust CUSUM Control Charting for Process Dispersion

Process monitoring through control charts is a quite popular practice in statistical process control. From a statistical point of view, a superior control chart is one that has an efficient design structure, but having resistance against unusual situations is of more practical importance. To have a compromise between the statistical and practical purposes, a natural desire is to have a control chart that can serve both purposes simultaneously in a good capacity. This study is planned for the same objective focusing on monitoring the dispersion parameter by using a Cumulative Sum (CUSUM) control chart scheme. We investigate the properties of the design structure of different control charts based on some already existing estimators as well as some new robust dispersion estimators. By evaluating the performance of these estimators‐based CUSUM control charts in terms of average run length, we identify those charts that are more capable to make a good compromise between the aforementioned purposes in terms of statistical and practical needs. Copyright © 2013 John Wiley & Sons, Ltd.     

A new dual CUSUM mean chart

The conventional cumulative sum (CUSUM) chart is usually designed based on a known shift size. In usual practice, shift size is often unknown and can be assumed to vary within an interval. With such a range of shift size, the dual CUSUM (DCUSUM) chart provides more sensitivity than the CUSUM chart. In this paper, we propose dual Crosier CUSUM (DCCUSUM) charts with and without fast initial response features to efficiently monitor the infrequent changes in the mean of a normally distributed process. Monte Carlo simulations are used to compute the run length characteristics of one‐sided and two‐sided DCCUSUM charts. These run length characteristics are compared with those of the CUSUM, Crosier CUSUM, Shewhart‐CUSUM, and DCUSUM charts in terms of the integral relative average run length. It turns out that the proposed chart shows better performance when detecting a range of mean shift sizes. A real dataset is considered to illustrate the implementation of existing and proposed charts.     

CUSUM Schemes for Monitoring Prespecified Changes in Linear Profiles

Because of the characteristics of a system or process, several prespecified changes may happen in some statistical process control applications. Thus, one possible and challenging problem in profile monitoring is detecting changes away from the ‘normal’ profile toward one of several prespecified ‘bad’ profiles. In this article, to monitor the prespecified changes in linear profiles, two two‐sided cumulative sum (CUSUM) schemes are proposed based on Student's t‐statistic, which use two separate statistics and a single statistic, respectively. Simulation results show that the CUSUM scheme with a single statistic uniformly outperforms that with two separate statistics. Besides, both CUSUM schemes perform better than alternative methods in detecting small shifts in prespecified changes, and become comparable on detecting moderate or large shifts when the number of observations in each profile is large. To overcome the weakness in the proposed CUSUM methods, two modified CUSUM schemes are developed using z‐statistic and studied when the in‐control parameters are estimated. Simulation results indicate that the modified CUSUM chart with a single charting statistic slightly outperforms that with two separate statistics in terms of the average run length and its standard deviation. Finally, illustrative examples indicate that the CUSUM schemes are effective. Copyright © 2016 John Wiley & Sons, Ltd.     

Dynamic Probability Control Limits for Lower and Two‐Sided Risk‐Adjusted Bernoulli CUSUM Charts

Because of its advantages of design, performance, and effectiveness in reducing the effect of patients' prior risks, the risk‐adjusted Bernoulli cumulative sum (CUSUM) chart is widely applied to monitor clinical and surgical outcome performance. In practice, it is beneficial to obtain evidence of improved surgical performance using the lower risk‐adjusted Bernoulli CUSUM charts. However, it had been shown that the in‐control performance of the charts with constant control limits varies considerably for different patient populations. In our study, we apply the dynamic probability control limits (DPCLs) developed for the upper risk‐adjusted Bernoulli CUSUM charts to the lower and two‐sided charts and examine their in‐control performance. The simulation results demonstrate that the in‐control performance of the lower risk‐adjusted Bernoulli CUSUM charts with DPCLs can be controlled for different patient populations, because these limits are determined for each specific sequence of patients. In addition, practitioners could also run upper and lower risk‐adjusted Bernoulli CUSUM charts with DPCLs side by side simultaneously and obtain desired in‐control performance for the two‐sided chart for any particular sequence of patients for a surgeon or hospital.