Efficient Dynamic Parameter Identification Framework for Machine Tools

Abstract

Modeling the dynamic behavior of a machine tool accurately is a difficult but crucial task when optimizing a machine tool’s design. An accurate representation of the real behavior is essential to ensure the transferability of simulations from a virtual prototype to a physical prototype. Due to the complexity of modern machine tools, a large number of parameters have an influence on the dynamic behavior. The parameterization of the used dynamic models is still challenging, especially if intricate local models are used for the individual effects. This paper presents an efficient framework for parameterizing a dynamic model of a machine tool containing linear local damping and stiffness parameters. For parameter identification, measurements of single components on simple test rigs as well as measurements of the whole machine tool were carried out. Different numerical optimization algorithms as well as objective functions were compared and applied to a three-axis machine tool structure for parameter fitting. By using a parametric reduced-order flexible multibody model for the fitting, high accuracy can be combined with high computational efficiency. The use of the presented approach allows an efficient parameter estimation and lays the groundwork for an influence analysis and the targeted optimization of a machine tool.