A controller-neutral representation for deterministic translation and regeneration of cnc programs across heterogeneous controllers
摘要
The interoperability of CNC systems remains a persistent challenge in modern manufacturing, as CNC programs are tightly coupled to controller-specific dialects, limiting their reuse across heterogeneous machine environments. Existing approaches, including CAM post-processing, STEP-NC-based workflows, and toolpath reconstruction methods, either depend on upstream CAD/CAM data or focus primarily on geometric interpretation, thereby limiting the systematic regeneration of legacy CNC programs across different controllers. This work reformulates CNC program translation as a semantic reconstruction problem and introduces a Controller-Neutral CNC Representation (CNCR) that captures machining semantics directly from executable CNC programs. The proposed representation decouples controller-specific syntax from machining intent by modeling modal machine states, machining actions, and normalized toolpath primitives within a unified intermediate framework. A state-driven interpretation process reconstructs implicit execution context, enabling deterministic regeneration—under reconstructed modal states—of functionally equivalent CNC programs for different target controllers without requiring access to the original CAM environment. The framework is validated through multiple industrial case studies involving heterogeneous controllers, including Fanuc, Siemens, Heidenhain, and Haas systems. Validation is performed at both code and geometric levels using simulation and quantitative toolpath comparison metrics, demonstrating deviations within typical machining tolerances and preservation of operation sequencing. The results demonstrate that the proposed approach enables controller-independent translation of legacy CNC programs while preserving machining intent, operational consistency, and geometric behavior. Overall, the framework provides a practical and systematic solution for CNC interoperability by bridging low-level CNC code with a controller-independent semantic representation.