FORM TOLERANCE ESTIMATION USING JACKKNIFE METHODS

A coordinate measuring machine (CMM) is a computer controlled device that uses a programmable probe to obtain measurements on a part surface. Recently CMMs have become very popular for dimensional measurement in industry due to their flexibility, accuracy, and ease of automation. Despite the advanta...

Full description

Saved in:
Bibliographic Details
Published in:Statistica Sinica 1998-01, Vol.8 (1), p.119-133
Main Authors: Shao, Jun, Tsui, Kwok-Leung
Format: Article
Language:English
Subjects:
Consistent estimators
Coordinate systems
Estimation bias
Estimation methods
Estimators
Jackknife resampling
Least squares
Quality and Reliability Improvement
Sample size
Sampling bias
Sampling methods
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A coordinate measuring machine (CMM) is a computer controlled device that uses a programmable probe to obtain measurements on a part surface. Recently CMMs have become very popular for dimensional measurement in industry due to their flexibility, accuracy, and ease of automation. Despite the advantages offered by CMM's, problems have emerged with their use because tolerance standards require knowledge of the entire surface while a CMM provides only a sample of points on the surface. These problems could be quite challenging, and both practitioners and researchers have shown great interest. Among these problems, estimating form tolerances for different part features is very important to practitioners. The least squares and minimum zone methods are the most commonly used methods for form tolerance estimation. Dowling et al. (1995) show that these two methods give seriously biased estimates of the part deviation range when the sample size is small. This paper establishes the consistency of these two common estimates. We propose several jackknife estimates that correct the bias of the least squares and minimum zone estimates. Based on a simulation study, it is found that the jackknife estimates effectively reduce the bias of the two common estimates in many situations, and thus reduce the chance of accepting bad parts in tolerance verification. We also show that the jackknife estimates are consistent.
ISSN:1017-0405
1996-8507