In-process digital twin estimation for high-performance machine tools with coupled multibody dynamics

Direct drive rotary and linear actuators significantly enhance the performance of multi-axis machine tools. The absence of mechanical gearing, however, increases the nonlinear dynamic coupling between the axes, making it challenging to identify accurate virtual models or so-called ‘digital twins’. This article presents a new approach to estimate nonlinear multivariable dynamic models non-intrusively, using in-process CNC data. Major influences, such as multi-rigid body motion, actuator force/torque ripples, nonlinear friction, feedforward/feedback control, and vibration modes, are systematically detected and identified. The new method is demonstrated in digital twin estimation for a 5-axis laser drilling machine.