Adaptive accuracy improvement of machine tools

Placid M Ferreira, Purdue University

Abstract

In this thesis, the problem of controlling the errors caused by the machine is addressed. The quasistatic errors are identified as the major source of dimensional errors and a practical approach for an automated machine shop floor is developed.^ The thesis describes various sources of quasistatic errors and methods used to control them. The problem of modeling the geometric errors of a machine and periodically updating this model is identified as being central to the strategy for controlling the quasistatic error.^ This model is developed by using rigid body kinematics. Shape and Joint transformations are developed for inaccurate links and joints(axes). The kinematic equations for a three-axis machine are then solved, assuming linear error characteristics for its joints.^ The problem of applying this model to the compensation of errors of a NC machine working in a manufacture environment is addressed. To accomplish this, the problem of determining the model's parameters by a simple procedure which can be executed between work cycles of the machine is found to be essential. It is then shown that, when the positioning errors of the axes are removed from the error model, the rest of the geometric error components can be determined by simple linear measurements at nine reference points in the workspace. This proves that it is possible to update the model's parameters at regular intervals, and possible to compensate the errors caused by the quasistatic effects.^ Finally, this thesis contains the experimental verification of the error model and the updating procedure. Using touch-trigger probes and a reference frame, the errors across a 2-D section of the workspace are predicted. The comparison of the predicted and observed errors proves conclusively the effectiveness of the model and the updating procedure. An order of magnitude improvement was observed in the locational accuracy across the 2-D workspace.^ The geometric error model developed in this thesis, besides its obvious application in error compensation, can be used in the selection of members with matching error characteristics at the design stage of the machine to improve the expected accuracy. Extensions to 5-axis machines and industrial robots are also possible. (Abstract shortened with permission of author.) ^

Degree

Ph.D.

Advisors

Major Professor: C. R. Liu, Purdue University.

Subject Area

Engineering, Industrial

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS