A NEURAL NETWORK MODEL 0OF A CMM APPLIED FOR MEASUREMENT ACCURACY ASSESSMENT

J. Sladek

Abstract

The paper presents research on CMM virtual modelling applied for assessment of measurement accuracy and a method of CMM errors identification. The idea of the proposed method of error estimation is based on the measurement of a workpiece plate (hole or ball), placed in the CMM measuring area in such way, that reference points compose a spatial grid. The difference between co-ordinates of particular shape elements midpoints obtained from workpiece calibration and the coordinates given by the CMM creates the error grid. This grid is a basis for a matrix method of CMM error identification. The identification matrix corresponds to the reference points distribution. The matrix model for the CMM error identification is composed of two component parts: one - CMM errors depending on the position in the measuring area of the tested machine and the other - independent of this position. An idea of a virtual model is based on artificial neural networks. Results of comparative research into various virtual models of measuring machines have been discussed.

Keywords

CMM, neural network, virtual model

Download

IMEKO-WC-2000-TC11-P295.pdf

DOI

-

IMEKO TC

TC11 - Measurement in Testing, Inspection and Certification

Back to the proceedings