Abstract
Performance evaluation of computer numerical control (CNC) machine tools is challenging due to high costs, experimental difficulties, and data scarcity. Digital twin technology can effectively address this issue through interactions between the physical and virtual environments. An executable high-fidelity digital twin is the foundation for utilizing digital twin technology in performance evaluation. Current research rarely focuses on the simplification methods and fidelity evaluation of CNC machine tool digital twins. Therefore, this paper proposes a novel digital twin construction methodology and introduces fidelity evaluation metrics based on the three-layer division of meta-action theory. Additionally, a multi-indicator weight allocation method is presented for the quantitative analysis of model updates and iterations. Firstly, CNC machine tools are divided into three layers: system level, chain level, and unit level based on meta-action theory. The digital twin architecture comprises the physical entity, information assurance layer, virtual entity, and functional application layer. Furthermore, a multi-level assessment framework is established, incorporating general indicators, system-level indicators, and unit-level indicators. To facilitate a qualitative and quantitative evaluation of the digital twins, the analytic hierarchy process and entropy weighting method are employed for weight allocation. This proposed approach simplifies the construction of digital twins and provides robust evaluation methods for their enhancement. The effectiveness of the proposed methodology is validated through a case study on the axial stiffness performance testing of a specific CNC machine tool rotary table model.