Task Coordinate Frame Based Contouring Control For Machine Tool Feed Systems

For high speed machines, e.g. CNC machine tool, high speed and high accuracycontouring control for multi-axis feed system is a crucial problem. Traditional controlleris designed based on the tracking error. However, the reduction of tracking error doesn’talways lead to the improvement of the final work piece. The contouring error isintroduced in the contouring control, which has been a research hotspot domestic andoverseas in recent years.In high speed and large curvature contouring applications, Task Coordinate Frame(TCF) based contouring control method lacks the capability to obtain the accuratecontouring error estimation and the contouring performance gets worse with theincrease of feed-rate. In order to settle this problem, the Task Polar Coordinate Frame(TPCF) based decoupling contouring controller is proposed in this dissertation. TheTPCF is established based on the circular approximation, which has a better contouringerror estimation compared with the tangent line approximation. The tracking error isdecomposed into two orthogonal parts, tangential error (along the angular direction) andradial error, and the last part is treated as the estimated contouring error. By thefeedback linearization and feed-forward compensation based on the reference input andidentified system dynamics, an error dynamics system is set up in the TPCF. Thecontouring performance is improved by enhancing the radial tracking capability. Variousexperiments on an XY-stage with typical contours, a circle, an ellipse, and a figure-’8’,were conducted. The contouring errors were dramatically reduced (above40%, RMS,the same in the followings) compared with the TCF based method, especially in highspeed and large curvature contouring cases.Existing tangent line approximation and circular approximation are both based onthe local geometry properties of the reference contour, and can be classified as thefirst-order and quasi-second-order local approximations. Besides, the circularapproximation is hard to be extended to the spatial case. A higher order localapproximation–natural local approximation is proposed in this dissertation, which isthird-order for the spatial contours and second-order for the planar ones. The classicalTCF based contouring controller is improved based the natural approximationcompensation. Simulink simulation of a simple helix contour verified that about38%contouring errors had been reduced. The ellipse contouring experiments on the XYstable indicated that the proposed method can decrease above60%contouring errorcompared with tangent line approximation and the contouring performance was almostthe same as the circular one (about2%reduction).The natural approximation is also utilized to improve the Position-Loop Cross-Coupled Controller (PLCCC). The PLCCC with a simple controller structure andstrong robustness has a good engineering application prospects. Ellipse simulationcontouring experiment indicated that the developed PLCCC adopted the naturalapproximation can decrease the contouring error by7%(RMS) and23.3%(Max). Thecontouring performance under the high speed and large curvature case has beenimproved.