Autonomous Thermal Management in CNC Machine Tools using Edge-Based Temporal Fusion Transformers and DDPG

Thermal management in Computer Numerical Control (CNC) machine tools is critical for enhanced operational integrity and tool longevity. Existing methods address thermal challenges reactively, causing potential inefficiencies and downtime. This study introduces an autonomous thermal management system using Edge-Based Temporal Fusion Transformers (TFT) and Deep Deterministic Policy Gradient (DDPG) algorithms. TFT captures complex temporal patterns from historical and real-time CNC thermal data to forecast temperature trends, enabling accurate predictions of potential overheating and other thermal failures, while DDPG utilises the TFT predictions to dynamically adjust CNC cutting parameters and cooling system settings to maintain optimal thermal conditions. The TFT-DDPG integration on edge-based architecture facilitates decentralised low-latency processing, allowing rapid response to prevent potential thermal failures. Experimental results on a CNC lathe showcase improved thermal stability, precision, and reduced tool wear. This study advances autonomous manufacturing by demonstrating the potential of edge-based intelligent systems for proactive thermal management.