An Adaptive Hybrid Edge-Cloud Collaborative Offloading Method for Large-Scale Computational Tasks of Intelligent Machine Tool: Low-Latency, Energy-Efficient, and Secure

Intelligent machine tools operating in continuous machining environments are commonly influenced by the coupled effects of multi-component degradation and updates in machining tasks. These factors result in the generation of vast multi-source sensor data streams and numerous computational tasks with interdependent data relationships. The stringent real-time constraints and intricate dependency structures present considerable challenges to traditional single-mode computational frameworks. Furthermore, there is a growing demand for computational offloading solutions in intelligent machine tools that extend beyond merely optimizing latency. These solutions must also address energy management for sustainable manufacturing and ensure security to protect sensitive industrial data. This paper introduces an adaptive hybrid edge-cloud collaborative offloading mechanism that combines single-edge-cloud collaboration with multi-edge-cloud collaboration. This mechanism is capable of dynamically switching between collaborative modes based on the status of computational nodes, task characteristics, dependency complexity, and resource availability, ultimately facilitating low-latency, energy-efficient, and secure task processing. A novel hybrid hyper-heuristic algorithm has been developed to address large-scale task allocation challenges in heterogeneous edge-cloud environments, enabling the flexible allocation of computational resources and performance optimization. Extensive experiments indicate that the proposed approach achieves average enhancements of 27.36% in task processing time and 7.89% in energy efficiency when compared to state-of-the-art techniques, all while maintaining superior security performance. Validation through case studies on a digital twin gantry five-axis machining center illustrates that the mechanism effectively coordinates task execution across multi-source concurrent data processing, complex dependency task collaboration, high-computational machine learning workloads, and continuous batch task deployment scenarios, achieving a 37.03% reduction in latency and a 25.93% optimization in energy use relative to previous generation collaboration methods. These results provide both theoretical and technical backing for sustainable and secure computational offloading in intelligent machine tools, thereby contributing to the evolution of next-generation smart manufacturing systems.