ISSN:0022-4065    

DOI:10.1080/00224065.1994.11979536

出版商:Taylor&Francis    

年代:1994

数据来源: Taylor

摘要:

Taguchi's experimental system is used to develop robust processes, that is, processes which are insensitive to variations in uncontrollable variables, such as environmental variables. Taguchi's recommended designs can be considered to be response surface designs. These designs are compared to composite designs, which are classical response surface designs, and a new class of composite designs having mixed resolution is suggested as a competitor to Taguchi designs. An example is given of types of problems where it is better not to achieve a robust process, and customizing for specific customers can lead to greater customer satisfaction. An introduction to response surface methodology and to screening designs is used to set the stage.

ISSN:0022-4065    

DOI:10.1080/00224065.1994.11979537

出版商:Taylor&Francis    

年代:1994

数据来源: Taylor

摘要:

A major benefit of modern“up front”quality assurance is that, once process capability and stability are achieved, costly inspection of the final product from each lot can be replaced by audit sampling. In this case study the non-trivial practical and statistical considerations that arise in developing such an audit sampling scheme are described. Our application involves an extrusion process for an engineering plastic produced in lots. There are one-sided specifications on each of six physical properties. Traditionally, samples were taken from each lot. To replace this, an audit sampling procedure was developed. With audit sampling eachkthlot is sampled unless the results show the need to return to every-lot inspection. In addition, the audit sampling results are monitored statistically to detect changes. The underlying criteria, the multivariate statistical plan, and various practical issues are discussed in this paper.

ISSN:0022-4065    

DOI:10.1080/00224065.1994.11979538

出版商:Taylor&Francis    

年代:1994

数据来源: Taylor

摘要:

Individuals and moving range charts are often used when production volume is too low to justify rational subgrouping. This article considers several aspects of the design strategy for these charts. We have found that theXchart alone is nearly as efficient as the combinedX-MR chart for detecting changes in the process variability. We recommend, however, that the control limits on the individuals chart be based on a shortterm estimate of the process variability, such as the moving range, rather than on a long term estimate, such as the sample standard deviation of the process. We also consider the effects of nonnormality and autocorrelation.

ISSN:0022-4065    

DOI:10.1080/00224065.1994.11979539

出版商:Taylor&Francis    

年代:1994

数据来源: Taylor

摘要:

Small 2k-pdesigns are increasingly being used in industry for process and product experimentation. Blocking of these designs can often significantly increase their efficiency. To facilitate the practical use of blocking, a comprehensive table providing information about blocking generators for all the possible eight-run and 16-run, two-level fractional factorial designs is presented. Examples of blocking and application of the table are also provided.

ISSN:0022-4065    

DOI:10.1080/00224065.1994.11979540

出版商:Taylor&Francis    

年代:1994

数据来源: Taylor

摘要:

Graph-aided methods for accommodating the estimation of interactions in factorial experiments have become popular among industrial users. Notable among them is the method of linear graphs due to Taguchi. Previously, some shortcomings of Taguchi's linear graphs were pointed out and an alternative method was proposed. By extending this method some new graphs for three-level fractional factorial designs have been developed. The proposed graphs have two new features. First, each edge of the graph can have one or two lines representing the two components of interaction in a three-level design, and second, there are two types of vertices and lines. A collection of graphs is given for 27-run and 81-run designs.

ISSN:0022-4065    

DOI:10.1080/00224065.1994.11979541

出版商:Taylor&Francis    

年代:1994

数据来源: Taylor

摘要:

A FORTRAN program is given to calculate the run length distribution (RLD), the average run length (ARL), and the standard deviation of the run length (SDRL) for residual control charts used to monitor autocorrelated process output. RLD, ARL, and SDRL values are calculated for processes that can be modeled by pure autoregressive models of orderp(AR(p)), pure moving-average models of order 1 (MA(1)), and mixed autoregressive moving-average models of orderspand 1 (ARMA(p,1)), given that the assignable cause to be detected is a step shift in the process mean.

ISSN:0022-4065    

DOI:10.1080/00224065.1994.11979542

出版商:Taylor&Francis    

年代:1994

数据来源: Taylor

ISSN:0022-4065    

DOI:10.1080/00224065.1994.11979543

出版商:Taylor&Francis    

年代:1994

数据来源: Taylor

ISSN:0022-4065    

DOI:10.1080/00224065.1994.11979544

出版商:Taylor&Francis    

年代:1994

数据来源: Taylor

ISSN:0022-4065    

DOI:10.1080/00224065.1994.11979546

出版商:Taylor&Francis    

年代:1994

数据来源: Taylor