Multi-variable optimal numerical control using adaptive model for identification of thermally induced deformation in high-speed machine tools

• Main Author: Fraser, Steven
• Format: Others
• Published: 1997
• Online Access: http://spectrum.library.concordia.ca/298/1/NQ39792.pdf; Fraser, Steven <http://spectrum.library.concordia.ca/view/creators/Fraser=3ASteven=3A=3A.html> (1997) Multi-variable optimal numerical control using adaptive model for identification of thermally induced deformation in high-speed machine tools. PhD thesis, Concordia University.

Thermally induced deformation in machining has been accorded an equal place with other sources of error, namely tool wear and mechanical deflection. The demand of the manufacturing sector to control the residual thermal errors below ±10 om in the whole working range of the machine tool has not yet been achieved. Since it is impossible to design machine tools that are thermally stable within this range, a feedback control system emerges as a logical and practical solution. The major problem in implementing real-time control systems for thermal deflection is that it is not possible to directly measure the relative thermal displacement between the tool and the workpiece during machining, prohibiting the use of true feedback control. Process models relating the thermal deformation to the temperature rise at some points on the structure are frequently used, but since complicated models are not practical in a real-time control environment, simplified empirical models of the structure are employed. Based on the results reported in the literature, one can conclude that this will lead to poor predictions. In order to improve the accuracy and reliability of thermal deflection estimation, a new concept of generalized modeling is expanded in this thesis to develop accurate real-time process models relating practical measured temperature points to the net thermal deflection of a machine tool structure. The first of these models deals with the identification of the generation of heat sources from the delayed temperature time response of measured points on the structure. The second model deals with the identification of the nonlinear effects that are introduced by contact interfaces within the machine tool structure. It has been shown that the accuracy of these models is within ±4 microns for typical configurations of a multi-component machine tool. The estimated thermal deflection error is compensated by means of a feedback/feedforward control system, using NC position control and electronically-controlled resistance heating pads as a micro-positioning actuation mechanism. The feedforward controller was designed using the method of model inversion, and the feedback controller was optimized using the LQR error minimization technique. The procedure was validated on linear and nonlinear machine tool models. It has been shown that the estimation and control system reduces the thermal deflection error to within ±10 microns, a reduction of 96%, for multi-component machine tool structures of typical geometry, using economical hardware and data acquisition techniques