Direct correction of thermo-elastic deformations
Final Report Abstract
This project developed a new, cost-effective, machine-integrated measurement method for small and medium 3-axis milling machines to determine thermo-elastic defects. The measurement setup consists of a table-sided thermostable laser frame with pentaprisms that deflect the beam path along the three coordinate directions of the machine axes and a measuring head in the spindle with two integrated position-sensitive diodes. The measurement is a relative measurement, which performs a comparison to an initial reference measurement. A measurement model calculates 13 of the 21 relevant thermo-elastic errors from the recorded straightnesses. This includes the squarenesses, which are particularly important for the thermo-elastic behavior, and two positioning accuracies. Since the pitch errors, which are important for the Abbe errors, are not measured, the application is focused on small and mediumsized milling machines. Due to the shorter Abbe levers of these kinematics, the pitch errors have a smaller influence on the total error. A measurement uncertainty for each individual error was determined for the process. The measurement system was validated with a commercial PSD measurement device and other measurements. The first objective of the application, the characterization and validation of a machine-integrated measurement method based on the photoelectric effect for small and medium-sized milling machines, was thus achieved. If the measurements of the system cannot take place in non-productive times, they reduce the productive machine time. In order to keep these non-productive measurement times as small as possible, the measurement method was supplemented with a modeling of the environmental and axis load influence. This modeling calculates the displacement of the Tool Center Point (TCP) for the measurable defects based on machine-internal data as well as the temperatures of the glass scales and the environment. This approach requires only a small number of additional sensors, making it cost-effective and less prone to failure. Unscented Kalman filters continuously determine parameters for the internal system description of the measurable thermoelastic errors of the milling machine. The system is described using first order delay elements. The model is adaptive, since it later trains load scenarios not considered in the initial parameterization. Furthermore, it is suitable for non-linear systems. Experimentally, an estimation interval of two hours was determined for the model. Thus, a new measurement is needed every two hours to update the model parameters. This corresponds to a maximum reduction in productive machine time of 3.3%. A series of measurements over more than three weeks validates the effectiveness of the correction. The root mean square value of the measured errors was reduced by 64.8%, while the peak to valley value of the measured errors was reduced by 65%. The second objective of the application (characterization and validation of a correction method) was thus achieved.
Publications
-
Direct measurement of thermo-elastic errors of a machine tool. CIRP Annals, 70(1), 333-336.
Brecher, Christian; Spierling, Robert; Fey, Marcel & Neus, Stephan
-
MEASUREMENT UNCERTAINTY OF A NEW MACHINE-INTEGRATED, PSD BASED APPROACH FOR THERMO-ELASTIC ERROR MEASUREMENT. MM Science Journal, 2022(2), 5713-5720.
BRECHER, CHRISTIAN; SPIERLING, ROBERT & FEY, MARCEL
-
Learning model for the thermo-elastic machine tool behavior. In: RWTH Publications, 2023
Spierling, R.; Brecher, C.; Plum, F. & Fey, M.
