A New Grey Box Approach for Friction Modelling of Machine Tool Drives

Measurement of the process force in milling is usually conducted by using piezo-electric dynamometers which are costly and reduce the stiffness of the system. A less invasive alternative is an indirect estimation of cutting forces based on the power of the servo drives. However, a correction of frictional effects from the transmission system is necessary to achieve accurate results. Most machine tools are equipped with ball-screw drives whose friction behavior is highly nonlinear due to dependency on both velocity and position. This study provides a novel approach to consider all frictional and inertial effects in transmission behavior of ball-screw drives by utilizing the well-established generalized M AXWELL slip (GMS) model and combines it with a data-based approach, namely support vector regression (SVR). The approach acquires the internal states of the GMS model and uses them as an addition-nal input for the SVR. The model is validated on different experiments conducted on a five-axis machining center and compared to established friction models, as well as a sole SVR. The results show the model to have errors between 7% and 12% over the full working range of the x-and y-axes, respectively, outperforming the benchmark models significantly.