摘要:

The effect of a planned process change on product performance is to be evaluated by implementing the change directly on the manufacturing floor and assessing the results. It is shown that, instead of making a one-time change and comparing the results after the change with those before, a statistically meaningful evaluation generally requires“back and forth”comparisons. The exact manner in which this is done will depend upon practical considerations. The concepts are illustrated by an example dealing with the manufacture of an electronic component, where the claim that a process change would lead to a reduction in cost without an appreciable decline in quality had to be verified. Sample size requirements, based upon an initial determination and a sequential decision plan, are also described.

ISSN:0022-4065    

DOI:10.1080/00224065.1982.11978778

出版商:Taylor&Francis    

年代:1982

数据来源: Taylor

摘要:

Bayesian acceptance sampling plans, including the increasingly popular economically-based versions, require the specification of a prior distribution of defectives from lot to lot. This paper examines the relationship between defectives in the sample and defectives in the rest of the lot for each of five prior distributions. It is seen that the use of a binomial prior renders sampling useless and inappropriate.

ISSN:0022-4065    

DOI:10.1080/00224065.1982.11978779

出版商:Taylor&Francis    

年代:1982

数据来源: Taylor

摘要:

Regression analysis of data obtained in statistically designed experiments has become an important source of information for the quality practitioner working on improving industrial processes. Least squares regression analysis is the usual method used to fit an equation to data. Recently developed robust regression methods in many cases provide a more accurate equation and these methods can be performed using available statistical software. Cases where robust methods would provide a more accurate equation are described and the reality of these situations in industrial or laboratory experiments is discussed. The value and practical aspects of using robust regression are illustrated with data from real experiments.

ISSN:0022-4065    

DOI:10.1080/00224065.1982.11978780

出版商:Taylor&Francis    

年代:1982

数据来源: Taylor

摘要:

Military Standard 105D has a provision for discontinuing inspection and stopping shipment of a product when evidence of poor quality is very strong. Discontinuation is not always feasible for critical products. In this paper, four rules are given for continued inspection of product during the discontinue phase of MIL-STD-105D. The rules allow up to ten more lots to be inspected using tightened sampling plans. Five criteria are associated with the four rules. An investigation of these criteria shows that the probability of satisfying them is large when the product has an historical quality equal to the AQL value and small when the historical quality is 2•AQL or worse. It is recommended that these rules be incorporated into the next revision of MIL-STD-105D.

ISSN:0022-4065    

DOI:10.1080/00224065.1982.11978781

出版商:Taylor&Francis    

年代:1982

数据来源: Taylor

摘要:

A computer program for the optimal economic design of an X̄control chart is presented. A single assignable cause system is assumed, where the mean time between process shifts is an exponentially distributed random variable. Given fixed and variable sampling costs, the costs of investigating action signals, the penalty cost of production in the out-of-control state, and other parameters describing process performance, the program finds the sample size, control limit width and interval between samples that minimize the expected total costs per unit time.

ISSN:0022-4065    

DOI:10.1080/00224065.1982.11978782

出版商:Taylor&Francis    

年代:1982

数据来源: Taylor

ISSN:0022-4065    

DOI:10.1080/00224065.1982.11978783

出版商:Taylor&Francis    

年代:1982

数据来源: Taylor

ISSN:0022-4065    

DOI:10.1080/00224065.1982.11978784

出版商:Taylor&Francis    

年代:1982

数据来源: Taylor

ISSN:0022-4065    

DOI:10.1080/00224065.1982.11978785

出版商:Taylor&Francis    

年代:1982

数据来源: Taylor

ISSN:0022-4065    

DOI:10.1080/00224065.1982.11978787

出版商:Taylor&Francis    

年代:1982

数据来源: Taylor

ISSN:0022-4065    

DOI:10.1080/00224065.1982.11978788

出版商:Taylor&Francis    

年代:1982

数据来源: Taylor