Optimization of characteristic diagram based thermal error compensation via load case dependent model updates

The compensation of thermal errors in machine tools is one of the major challenges in ensuring positioning accuracy during cutting operations. There are numerous methods for both the model-based estimation of the thermal tool center point (TCP) deflection and for controlling the thermal or thermo-elastic behavior of the machine tool. One branch of thermal error estimation uses regression models to map temperature sensors directly onto the TCP-displacement. This can, e.g., be accomplished using linear models, artificial neural networks or characteristic diagrams. One of the main limitations of these models is the poor extrapolation behavior with regard to untrained load cases. This paper presents a new method for updating characteristic diagram based compensation models by combining existing models with new measurements. This allows the optimization of the compensation for serial production load cases without the effort of computing a new model. The new method was validated on a 5-axis machining center.