Lightweight design of production machines results in a problematically high inclination of vibration. Existing control techniques can avoid those vibrations based on the knowledge of the instantaneous system behavior, which is highly non-linear and dependent on the machine’s pose. One possibility is to calculate the machine behavior using a model of the machine on the control device. This requires a real-time capable algorithm of system behavior identification calculated within the control device cycle. In this paper we describe a new approach for model-based system behavior calculation in real-time on the control device.