Identification and control of high-speed machine tools

The drive to increase manufacturing productivity has created a demand for high-performance, high-bandwidth mechanical systems. This dissertation constructs the theoretical framework for the development of such systems, specifically focusing on the design of a controller for a high-speed milling machine.; A feedforward control architecture is presented for feed drive control which combines the Zero Phase Error Tracking Controller (ZPETC) and feedforward friction compensation. The ZPETC is designed using a closed-loop plant model obtained from a novel identification procedure, frequency-weighted least squares. This methodology is able to produce a transfer function model of modest order which can accurately identify the low frequency dynamics of a system. The feedforward friction compensation is derived from the converged outputs of a repetitive controller. Repetitive control is able to achieve near-perfect tracking of a reference trajectory in the presence of low velocity friction given a dozen or less learning passes. The nonlinear control commands which achieve this performance are incorporated in a look-up table and are applied as feedforward input according to the desired command velocity.; Tracking and contouring experiments are performed on the X-Y bed of the machine tool using the proposed feedforward control architecture. The tests demonstrate that near-perfect tracking and contouring (within {dollar}pm{dollar}5 {dollar}mu{dollar}m) can be achieved for a variety of trajectories at feedrates up to 3.0 m/min. In one experiment at 10.0 m/min, a maximum contouring error under 11 {dollar}mu{dollar}m is obtained for a circular profile. End milling experiments are conducted for three cutting geometries: circular radial machining, rounded corner cutting, and boring. The maximum spindle speed of 15,000 RPM is used with feedrates of 2.0-3.0 m/min. The specimens machined using the feedforward control architecture are compared with those machined using an industrial CNC controller through coordinate measuring machine (CMM) data. The CMM measurements reveal that the feedforward control architecture is able to increase the bandwidth of the system while eliminating the effects of friction, thus removing the machining errors attributable to dynamic positioning. For one particular specimen, the experimental controller is able to reduce the machining error by almost a factor of 20.