Etalonnage geometrique des machines-outils par barre a billes en vue de predire leur performance (French and English text)

The development of a procedure to identify the geometric errors on a multi-axis machine tool is the main objective pursued in this research. Once the errors are identified, predicting the performance of the machine tool for a machining operation on a surface is the second objective of this work.; The geometric inaccuracies of a machine tool strongly influence the quality of machined parts. A systematic approach is developed for the calibration of the position independent geometric error parameters (PIGEPs) such as the joint misalignments of a five-axis machine tool. A kinematics model of such a machine tool is developed using the homogenous transformation matrix and incorporates geometric errors. This machine tool model was used to generate the virtual measurements, observation matrix and error prediction procedure. An observation matrix is developed based on the machine tool nominal model, configurations and topology. This matrix is used to calculate the sensitivity of the tool-location error to the geometric error sources. Then a telescoping magnetic ball bar (TMBB) is used to gather length readings for a number of machine configurations. From an error identification procedure and through a strategy of selection of error parameters, a set of error parameters are selected which represent a complete and minimal set of error parameters for a five axis machine tool. A series of criteria is presented for the acceptance of the identified parameters. Finally several tests are presented to validate the method through simulation. These show excellent results both for the parameter values and for the tool versus workpiece error prediction.; As a further step, a procedure is developed to calibrate a five-axis machine tool for position dependent geometric errors parameters (PDGEPs) again using the telescoping ball-bar. These PDGEP describe the non-ideal motion of joints. The model is based on the development of the PDGEP identification and uses polynomial functions for modelling the PDGEP error model. The machine tool error sources are individually modelled by Chebychev polynomials and integrated in the observation matrix. The Chebychev polynomials are used because of their advantages for computing purposes. The coefficients of these polynomials are calculated through a set of linear equations made of the observation matrix and ball-bar readings. Then, the identified polynomials are used to predict the tool-tip error for a given joint coordinate values. Finally, some criteria are used to evaluate whether the identified parameters are valid. The Simulation of the entire calibration procedure are developed based on Matlab™. (Abstract shortened by UMI.)...