Online parameter identification for a linear parameter-varying model of large-scale lightweight machine tool structures with pose-dependent dynamic behavior

Abstract

Structural components of large lightweight machine tools with serial kinematics, travelling column and Gantry-type machine tools, feature a pose-dependent dynamic behavior. Conducting vibration reduction methods at these structures demands information on the precise current dynamic behavior. Determining this, an approach is presented, where the parameters of a simple rigid multi-body model of a lightweight travelling column machine tool structure are adapted online to the current dynamic behavior of the structure using a recursive least-squares estimator. Inherent control signals and additional acceleration sensor signals are used for the parameter updating. These signals are conditioned using Kalman and lowpass filters. The model as well as the online parameter identification algorithms are validated at a laboratory prototype of a lightweight travelling column machine tool. Experiments show that the model parameters quickly converge to a stable state after impulse excitation of the laboratory prototype with fixed axis. The parametrized model exactly represents the measured dynamic behavior of the laboratory prototype in a frequency range until 50 Hz. For moving axis the estimated parameters consistently change according to the movement of the axis.