To accurately predict the precision degradation of CNC machine tools and ensure machining quality, we propose a precision prediction method based on "function-motion-action". The method starts from meta-action unit components. It quantifies error propagation using both adhesive wear and abrasive wear models. We establish accuracy degradation functions by analyzing geometric and kinematic relationships. The machining accuracy evolution patterns are derived through multi-body system theory and homogeneous coordinate transformation matrices. This method achieves reliable accuracy prediction without requiring actual machining data. In experiments on a Nanjing CNC lathe, the mean prediction error was 4 μm. This result demonstrates better interpretability and accuracy than conventional methods. This method offers a theoretical foundation and practical solution for maintaining CNC machine tool accuracy.