A computational algorithm for selecting robust designs in safety and quality critical processes

This article considers robust designs for safety and quality critical processes. In critical processes, the choice of control settings should ensure that the product quality remains near the target for each and every value of the noise variable. This objective is realized by using a min-max approach...

Full description

Saved in:
Bibliographic Details
Published inSankhyā. Series B (2008) Vol. 73; no. 1; pp. 105 - 122
Main Authors Mukhopadhyay, S., Chakraborty, D.
Format Journal Article
LanguageEnglish
Published India Springer 01.05.2011
Springer-Verlag
Indian Statistical Institute
Subjects
Algorithms
Analysis of variance
Automobiles
Control variables
Entropy
Error rates
Manufacturing processes
Mathematics and Statistics
Modeling
Noise
Noise control
P values
Product quality
Robust parameter design
Samkhya
Statistics
Drug granulation
62P30
62K25
Computational algorithms
Guaranteed quality
Crash test
Control setting
62K20
Optimization
Online AccessGet full text

Cover

More Information
Summary:This article considers robust designs for safety and quality critical processes. In critical processes, the choice of control settings should ensure that the product quality remains near the target for each and every value of the noise variable. This objective is realized by using a min-max approach which minimizes the maximum possible deviation (caused by noise) of the estimated response from the target value. An algorithm for computing such control settings based on entropic regularization is discussed. The proposed method is used on automobile crash test data to select maximally safe designs for the interior rim of cars. A second example on the design of drug granulation parameters to produce uniformly sized granules is also included.
ISSN:0976-8386
0976-8394
DOI:10.1007/s13571-011-0021-0