Hybrid modeling of transient volumetric machine tool errors for virtual climatization

  • Hybride Modellierung transienter volumetrischer Werkzeugmaschinenfehler für virtuelle Klimatisierung

Dahlem, Jan Philipp; Schmitt, Robert H. (Thesis advisor); Mayer, René (Thesis advisor)

Aachen : RWTH Aachen University (2023)
Dissertation / PhD Thesis

Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2023


The global trend towards more individual and complex products, and hence, smaller batch sizes, and tighter tolerances is ongoing. At the same time, the need for more sustainable production technologies and an overall smaller carbon footprint is higher than ever before. Industries are facing the challenge of meeting the resulting requirements of both developments. The reduction of the energy consumption of production companies is a central component of the international targeted transformation process. In 2020, 44.2 % of electricity consumption in Germany was attributable to industry. As machine tools are the backbone of modern production, their further development in terms of higher performance, increased volumetric accuracy in conjunction with significantly more efficient operation is essential.In order to ensure high accuracy requirements in production, complex disturbances due to temperature input are typically suppressed with energy-intensive air-conditioning and cooling measures. As an energy-saving alternative, the concept of virtual climatization for machine tools is presented in this thesis, which is only able to predict thermally transient errors of machine tools for a targeted compensation with the help of suitable models. The work is therefore concerned with the goal to identify and develop suitable models for this application and combining them into reliable hybrid models. Already known prior knowledge about the thermal behavior of machine tools shall be taken into account in the form of submodels. At the same time, the manual effort for the modeling shall be kept as low as possible to facilitate a broad application to different machine tool types. By means of a suitable approximation of physical structural deformations and a combination with machine learning algorithms, an efficient model adaptation to machine-specific conditions shall be implemented. In addition, the work investigated the question of how the necessary model data can be acquired for the model setup and also for the application in operation. A special focus is on the application of the overall concept as a retrofit solution for existing machines. The research work follows the method of Design Science Research. The developed artifacts, including the Abstracted Physical Body Model and the hybrid model structure for the combination with machine learning, as well as various concepts for the acquisition of suitable model data, are investigated and validated in a consolidated experiment, following the Design of Experiments format. A comparison of different combinations of submodels with respect to the model properties is carried out, whereby the advantages of the pursued approach can be clearly highlighted.


  • Laboratory for Machine Tools and Production Engineering (WZL) of RWTH Aachen University [417200]
  • Chair of Production Metrology and Quality Management [417510]