Implementation of Model Reference Adaptive Force Control in Milling.

Abstract

In the interest of maximizing the metal removal rate and preventing tool breakage in the milling process, fixed gain feedback controllers are used, which manipulate the feed rate to maintain a constant cutting force. These process controllers have resulted in substantial improvements in the metal removal rate; however, they may perform poorly when the process parameters deviate from the design conditions. One possible solution is the application of adaptive control. However, to apply an adaptive controller, a process model must be developed. Therefore, an empirical second order model of the force response to feedrate changes has been developed, and experimental results are presented which show that the parameters of this model vary significantly with cutting conditions. These variations are shown to have significant effects on the performance of fixed-gain controllers. This is demonstrated using machining tests as well as through digital simulations. A Model Reference Adaptive Controller has been designed and implemented on the milling system. The adaptive controller was found to perform more satisfactorily than the fixed gain controllers but is difficult to implement and tune, primarily because of the unmodeled dynamics resulting from runout on the milling cutter. Several alternatives were considered to solve the problems due to the unmodeled dynamics. Specifically, the output error method and the extended least squares estimation algorithms were considered and found to be unsatisfactory. In this problem there was sufficient separation between the noise and the signal frequency that the noise could be filtered. However, the addition of the filter added additional dynamics to the system which reduced the overall performance from that expected from the digital simulations.