Control-Oriented Modeling Requirements of a Direct-Drive Machine Tool Axis

Abstract

The high performance demands on commercial computer numerical control (CNC) machine tools have led to the widespread adoption of direct-drive servo axes. In industrial machines, where the workpiece is manipulated by the axis, the plant dynamics seen by the control system may vary widely between different workpieces. These changing plant dynamics have been observed to lead to limit-cycle behavior for a given controller. In such a situation, conventional modeling approximations used by practitioners may fail to predict the onset of instability for these axes. This work demonstrates the failure of conventional modeling approximations to predict the observed instability in an industrial CNC servo axis and investigates the model fidelity required to replicate the observations. This represents an important consideration when designing model-based controllers for direct-drive axes in CNC machines.