摘要:

Data from industrial experiments often involve an ordered categorical response, such as a qualitative rating. Analysis of variance-based analyses may be inappropriate for such data, suggesting the use of generalized linear models (GLM's). When the data are observed from a fractionated experiment, likelihood-based GLM estimates may be infinite, especially when factors have large effects. These difficulties are overcome with a Bayesian GLM, which is implemented via the Gibbs sampling algorithm. Techniques for modeling data and for subsequently using the identified model to optimize the process are outlined. An important advantage in the optimization stage is that uncertainty in the parameter estimates is accounted for in the model. For robust design experiments, the Bayesian approach easily incorporates the variability of the noise factors using the response modeling approach. This approach and its techniques are used to analyze two data sets, one of which arises from a robust design experiment.

ISSN:0040-1706    

DOI:10.1080/00401706.1996.10484411

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

For a binomial sampling model using pass/fail data and a beta prior distribution, parametric empirical Bayes (PEB) point and credibility interval estimates are provided for the time-and-diesel average emergency diesel generator (EDG) reliability to load-run on demand at 63 U.S. commercial nuclear power stations. For a geometric sampling model and a beta prior distribution, similar estimates of the time-dependent EDG-average reliability are also obtained for four plants. This is an expository article that introduces seven practical steps, involving both model validation and inference, in applying modern PEB methods. It also illustrates some important advantages of using industry-wide data to make inferences on individual items.

ISSN:0040-1706    

DOI:10.1080/00401706.1996.10484412

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

This article focuses on the problem of predicting future measurements on a statistical unit given past measurements on the same and other similar units. We introduce a conditional predictor that uses the information contained in previous measurements. The prediction technique is based on the iterative EM algorithm, but a noniterative variant is also provided. We use the sample-reuse methodology to select an appropriate predictor. The technique is illustrated in three engineering applications. The first considers prediction in the context of marketing for bucking in automatic forest harvesters. In fatigue-crack-growth data, the interest is in predicting the future crack-growth development of the test unit, and the third application concerns evaluation of pulp from the point of view of its papermaking potential.

ISSN:0040-1706    

DOI:10.1080/00401706.1996.10484413

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

Modern dimensional inspection techniques often produce more measurements per part than parts measured within reasonable timeframes. This poses a problem for multivariate process monitoring and capability analysis: The sample covariance matrix is not positive definite, hence not full rank and not invertible. When the measurement sites form a multidimensional lattice, spatially stationary covariance models provide positive definite estimates regardless of the number of measurements per part. I show that these estimates may be used in place of the sample covariance matrix to extend, and in some cases improve, standard multivariate methods. I describe a general class of lattices for which positive definite estimates are obtained via simple averaging or a closed-form EM algorithm. The proposed estimation and analysis procedures are illustrated in three case studies.

ISSN:0040-1706    

DOI:10.1080/00401706.1996.10484414

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

The performance of slapper detonators depends on the sensitivity of the explosive material within the detonators to a voltage stimulus. A series of experiments is typically used to estimate the sensitivity of the explosive material by subjecting individual detonators to various stimulus levels. The traditional methods for selecting stimulus levels and analyzing the resulting binary data are based on a parametric probit model that relates the probability of detonation to the level of the voltage stimulus. In this article, we explore an alternative method for analyzing the binary data usingnonparametricregression. The use of thisalternativemethod is illustrated byanalyzingthe ungrouped binary data resulting from a series of 25 tests that were used to characterize a single lot of detonators. We conclude that nonparametric regression can be very useful in visualizing the underlying relationship between a binary response and a covariate, even when the sample size is relatively small.

ISSN:0040-1706    

DOI:10.1080/00401706.1996.10484415

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

When indirect measurements are used to obtain rate-of-change data, an additional source of uncertainty is introduced that must be taken into account in the mean-rate and standard-error estimates. A procedure to account for this uncertainty is described. The chain rule of differentiation is used to decompose the rate into the product ofcalibration and primary experimentparts. Mean-rate and standard-error estimates are then obtained in terms of the component means and standard errors. The procedure is demonstrated by an analysis of some fatigue-crack-growth data.

ISSN:0040-1706    

DOI:10.1080/00401706.1996.10484416

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

Supersaturated designs are very cost-effective to scientists and engineers at the primary stage of scientific investigation. This article describes a method of constructing supersaturated designs from balanced incomplete block designs that is a generalization of the method of Lin for constructing these designs and a more general approach to constructing these designs.

ISSN:0040-1706    

DOI:10.1080/00401706.1996.10484417

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

ISSN:0040-1706    

DOI:10.1080/00401706.1996.10484418

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

ISSN:0040-1706    

DOI:10.1080/00401706.1996.10484419

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

ISSN:0040-1706    

DOI:10.1080/00401706.1996.10484420

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor