Load variation compensation in real-time for motion accuracy of machine tools

Although nanometric precision is nowadays possible for positioning tasks, it still not available for machining processes under cutting conditions because periodical forces produced by the cutting process on the work-table are neglected. This work proposes an approach to improve accuracy of conventional machine tools under cutting conditions for both positioning tasks and trajectory tracking tasks. A three axis-milling machine is used as experimental platform. A reduced second order linear model of a typical machine tool axis is considered. The main assumption is that cutting forces can be seen as load variations at the worktable level, these load variations are then considered as dynamic variations of the plant parameters. Using DSP and Pentium processors, a very fast identification process is able to estimate these variations in real-time, at the same rate as the servo-cycle.