Showing papers in "Journal of Quality Technology in 2009"
TL;DR: The past decade has seen rapid advances in the development of new methods for the design and analysis of split-plot experiments and the value of these designs for industrial experimentation has not been fully appreciated.
Abstract: The past decade has seen rapid advances in the development of new methods for the design and analysis of split-plot experiments. Unfortunately, the value of these designs for industrial experimentation has not been fully appreciated. In this paper, we review recent developments and provide guidelines for the use of split-plot designs in industrial applications.
213 citations
TL;DR: This work proposes the use of nonlinear mixed models to monitor the nonlinear profiles in order to account for the correlation structure and evaluates the effectiveness of fitting separate nonlinear regression models to each profile in Phase I control chart applications for data with uncorrelated errors and no random effects.
Abstract: Profile monitoring is a quality control technique best used when the process data follow a profile at each time period. Little work has been done on the monitoring of nonlinear profiles, and that work has assumed that the measurements within a profile a..
151 citations
TL;DR: A forward-selection algorithm is first utilized to screen out potential out-of-control variables; a multivariate control chart is then set up to monitor suspicious variables and detection of faulty conditions and isolation of faulty variables can be achieved in one step.
Abstract: A variable-selection-based multivariate statistical process control procedure is proposed for process monitoring and fault diagnosis in high-dimensional processes. Use of a forward-selection algorithm and a multivariate control chart enable in one step ..
150 citations
TL;DR: This paper showed that models based on engineering/physics laws don't often reflect reality and that statistical models commonly used in quality improvement studies often do not perform well when predictions are made away from the observed data points.
Abstract: Statistical models commonly used in quality improvement studies often do not perform well when predictions are made away from the observed data points. Yet models based on engineering/physics laws don’t often reflect reality. Engineering/statistical mod..
91 citations
TL;DR: A distribution-free method for defining the in- control state of a process and identifying an in-control reference sample is proposed and the proposed rank-based method is compared with the traditional X chart using Monte Carlo simulation and performs better than the X chart in many situations when the process distribution is skewed or heavy tailed.
Abstract: Statistical quality control often relies on the completion of a Phase I study. Many Phase I control charts, however, rely on an assumption of normally distributed process observations that may not be reasonable in the early stages of process control. A ..
90 citations
TL;DR: The complex data records collected by many modern industrial processes do not readily allow the use of traditional statistical process control techniques as discussed by the authors, and therefore, traditional statistical data structures do not work well in industrial processes.
Abstract: The complex data records collected by many modern industrial processes do not readily allow the use of traditional statistical process-control techniques An observation from a process might consist of n pairs of x and y data Such data structures or re
83 citations
TL;DR: In this article, the cumulative sum (CUSUM) control chart is investigated for monitoring autocorrelated processes of counts from a Poisson INAR(1) process.
Abstract: An attributes control chart generally assumes that the process being monitored is independent and identically distributed in its in-control state, but the chart’s performance may be degraded if the autocorrelation structure is not taken into account. The cumulative sum (CUSUM) control chart is investigated for monitoring autocorrelated processes of counts from a Poisson INAR(1) process. The new chart allows exact average run length computations. It is shown that the proposed CUSUM chart performs well in detecting assignable causes in autocorrelated count processes.
77 citations
TL;DR: In this paper, a robust Hotelling's T²-type control chart is proposed to monitor individual observations in a multivariate process, which is based on reweighted minimum covariance determinant estimators.
Abstract: A robust Hotelling’s T²-type control chart is proposed to monitor individual observations in a multivariate process. The chart is based on reweighted minimum covariance determinant estimators. It is shown how to set the control limits for the proposed c..
58 citations
Abstract: Inefficient utilization of operating rooms is a major problem in hospitals worldwide. A study of 13 hospitals in Belgium and the Netherlands showed that surgery began consistently late. Two hospitals were selected to record the start times for the first..
51 citations
TL;DR: An overview of statistical thinking and methods in the research and development and manufacturing functions in the pharmaceutical industry is given in this paper, where case studies illustrate how these issues work in real life settings.
Abstract: An overview is given of the use of statistical thinking and methods in the research and development and manufacturing functions in the pharmaceutical industry Four case studies illustrate how these issues work in real life settings A synopsis of these
48 citations
TL;DR: Two number-theoretic methods for generating space-filling designs for constrained mixture experiments defined by single- and multiple-component constraints are introduced.
Abstract: Two number-theoretic methods are introduced for generating space-filling designs for constrained mixture experiments defined by single- and multiple-component constraints. The methods are illustrated for a simple 3-component mixture problem and a more c..
TL;DR: In this article, a multivariate analysis tool is described that can detect change in the mean vector and/or covariance matrix in addition to the epoch of change in an independent sequence of multivariate observations.
Abstract: A multivariate-analysis tool is described that can detect change in the mean vector and/or covariance matrix in addition to the epoch of change in an independent sequence of multivariate observations. The multivariate change model is investigated using ..
TL;DR: This work uses Markov chain Monto Carlo methods to approximate properties of the posterior distributions of Bayesian models for models that are a compromise between the bad-as-old and the good- as-new models.
Abstract: Models for repairable systems can be characterized by the effect of the failure and the subsequent repair. Conventional repair models consider two extremes: As-bad-as-old models that lead to the nonhomogeneous Poisson process and the as-good-as-new mode..
TL;DR: The use of statistical methods have helped the advancement of nanotechnology as mentioned in this paper and have become a multidisciplinary subject that will affect every sector of the economy and daily life in the near future.
Abstract: Nanotechnology has become a multidisciplinary subject that will affect every sector of the economy and daily life in the near future. Science-based technologies and the use of statistical methods have helped the advancement of nanotechnology. Some insta..
TL;DR: In this article, it is assumed that the binary observations from many processes are autocorrelated, and this can have an adverse effect on the performance of the control chart, and it is shown that this can affect the performance.
Abstract: When monitoring a proportion ρ with conventional control charts, it is assumed that the binary observations are independent. However, the observations from many processes are autocorrelated, and this can have an adverse effect on the performance of the ..
TL;DR: In this paper, the mean and variability of a process has traditionally been done using separate X-bar and S charts, but it has been shown that it is possible to monitor both the Mean and Variability using a single chart responsive to shifts in either.
Abstract: Monitoring the mean and variability of a process has traditionally been done using separate X-bar and S charts. Several published studies have shown it is possible to monitor both the mean and variance using a single chart responsive to shifts in either..
TL;DR: A new acceptance/rejection rule suitable for use in start-up demonstration tests is introduced and study, according to which equipment is accepted if k consecutive successes occur, while the equipment is rejected if two failures separated by at most r — 2 successes are observed.
Abstract: An acceptance/rejection rule suitable for use in start-up demonstration tests is discussed. The rule allows acceptance of equipment if k consecutive successes occur, while the equipment is rejected if two failures separated by at most r – 2 successes ar..
TL;DR: In this article, the first occurrence of a pattern in a specific family of patterns occurring in a bivariate sequence of trinomial trials was investigated for the first time, and the exact distribution of the waiting time was derived using a Markov chain embedding technique.
Abstract: The waiting-time distribution is investigated for the first occurrence of a pattern in a specific family of patterns occurring in a bivariate sequence of trinomial trials. The approach can be applied in situations where the final quality of the process output is determined by two correlated variables used to classify the two measured characteristics. The exact distribution of the waiting time is derived using a Markov chain-embedding technique.
TL;DR: C constrained clustering is proposed as a practical, robust, general solution to detect change points and partition historical process data by modifying agglomerative clustering algorithms to take into account the time order within the data set.
Abstract: Standard statistical methods based on global models may not be effective for the analysis of massive multivariate industrial data sets unless the data is partitioned into stable operating regions. Constrained clustering is proposed as a robust solution to detect change points and partition historical process data. The constraint is that only observations that are contiguous in time can be joined. A method is described for partitioning data sets into stable regions by modifying agglomerative clustering algorithms to take into account the time order within the data set. A stopping criterion is proposed to evaluate the number of change points generated.
TL;DR: Compared with classical and robust estimation procedures, simulation studies show that the proposed computer-intensive multistep cluster-based method is usually better and sometimes much better at detecting randomly occurring outliers as well as outliers arising from shifts in the process location.
Abstract: The Hotelling’s T² approach for outlier detection in an individuals retrospective multivariate control chart gives unreliable results because the presence of outliers distorts both the sample mean vector and the covariance matrix. To remedy this problem, a computer-intensive, multistep cluster-based method is proposed to overcome the distortion or masking. Simulation studies show the procedure is better than classical or robust procedures at detecting randomly occurring outliers and outliers resulting from shifts in the process location. Comparisons based on real data are provided.
TL;DR: In this paper, Acceptance Sampling in Quality Control, 2nd edition, is discussed, with a focus on the acceptance sampling in the context of QC. But this paper is not a comprehensive review.
Abstract: (2009). Acceptance Sampling in Quality Control, 2nd edition. Journal of Quality Technology: Vol. 41, No. 4, pp. 441-442.
TL;DR: In this article, the information obtained at high stress levels in accelerated life tests is generally extrapolated with an appropriate statistical model to estimate parameters of interest at normal stress levels, however, these statistical analyses do not take batch di..
Abstract: The information obtained at high stress levels in accelerated life tests is generally extrapolated with an appropriate statistical model to estimate parameters of interest at normal stress levels. However, these statistical analyses do not take batch di..
TL;DR: Optimal settings of RM proportions and/or PV levels were developed that were less costly and more robust than the reference recipe and PV settings that had been tentatively identified by a nonstatistical approach.
Abstract: A process for producing low-fat mayonnaise investigates optimizing the raw material recipe and process variable levels. A mixture-process variable design was constructed and run as a split-plot experiment. Squared-error loss, mean-squared error, and boo..
TL;DR: The results of a process performance evaluation and process capability analysis was used to help identify and solve the problem of excess variability and indicated differences in quality between vendors and within vendors over time.
Abstract: Many service organizations today have problems with the quality of the credit card processing task. A case study shows the results of a process performance evaluation and process capability analysis used to identify and solve the problem of excess varia..
TL;DR: In this paper, a procedure for computing two types of errors that characterize split-plot experiments is described, i.e., R², R² adjusted, PRESS, and R²-pr.
Abstract: When determining measures of adequacy of fit for split-plot models, it is important to take into consideration the two types of errors that characterize split-plot experiments. A procedure is described for computing two R², R²-adjusted, PRESS, and R²-pr..
TL;DR: This work considers the assessment of an automated continuous measurement system used for 100% inspection in a high- volume manufacturing process, and describes an alternative sampling plan where parts with extreme values from the population of measured parts to remeasure.
Abstract: We consider the assessment of an automated continuous measurement system used for 100% inspection in a high- volume manufacturing process. Because of the automation, we assume that there are no operator effects. If the system stores the measured values, we effectively know the current process mean and standard deviation. Because of the high volume, we have parts available with values spread across the whole distribution. The standard plan for measurement-system assessment is to select k parts at random from the process and measure each of the selected parts n times. We then estimate the repeatability of the system using ANOVA. We propose two improvements. First, we demonstrate the substantial value of using the known process characteristics in the analysis. Second, we describe an alternative sampling plan where we deliberately select parts with extreme values from the population of measured parts to remeasure. We call this selection leveraging. We discuss the analysis of the leveraged plan and show that it is more efficient than the standard plan. We also discuss the planning and implementation of a leveraged assessment study and some associated issues and extensions.
TL;DR: Using this entropy model, the authors are able to reconcile the apparently random occurrence of accidents and errors with the observed systematic trend associated with a learning environment where the outcome distribution depends upon the depth of experience of the human participants.
TL;DR: In this paper, the experimental design used in this type of nonrepeatable gauge R&R studies is best constructed in a way that is similar to a Latin square design, and several examples in which this approach is applied and worked out are considered.
Abstract: The standard method to assess a measurement system’s precision is a gauge repeatability and reproducibility (gauge R&R) study. It exploits replications to estimate variance components that are interpreted as measurement spread. For nonrepeatable measurements, it is not feasible to obtain replications because objects are destroyed when they are measured or because the object changes over time. Possible solutions are to replace replications with measurements of multiple objects or with the measurement of one object at multiple times. Subsequently, these measurements are modeled by a fixed pattern (over time or over positions). We show that the experimental design used in this type of nonrepeatable gauge R&R studies is best constructed in a way that is similar to a Latin square design. These designs have a great flexibility, can be applied in many situations encountered in practice, and have nice mathematical properties as well. We consider several examples in which this approach is applied and worked out. For the examples given, we provide the analysis and the results following the worked-out approach. Analysis of the envisaged experimental set-up is done with linear and nonlinear mixed models in which variance components are estimated by restricted maximum-likelihood estimators.
TL;DR: In this paper, the authors provide helpful direction to the reader with over 450 figures, tables, and road maps, but these examples are not developed enough to adequately equip the reader to make the immediate improvements that are promised if he or she does not have prior knowledge or training.
Abstract: The over 450 figures, tables, and road maps do provide helpful direction to the reader. The figures and examples in the book usually are enough to communicate the concepts to the reader, but these examples are not developed enough to adequately equip the reader to make the immediate improvements that are promised if he or she does not have prior knowledge or training. The numerous tables that summarize much of the content in the book are helpful, especially as a quick reference or to be used as checklists when applying the tools or approaches presented. While these sometimes seem to simplify the tasks without mentioning the real life challenges in implementation, their concise nature is a benefit if not over-relied upon.