Multi-satellite assisted ocean-going ship ai task computing offloading method and system
By employing multi-satellite assistance and multi-agent reinforcement learning algorithms, the challenges of network planning and resource allocation in AI task computational offloading for ocean-going vessels were solved, resulting in an efficient and stable computational offloading scheme that enhances the intelligence level of ocean-going vessels.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- WUHAN SHIP COMM RES INST (NO 722 RES INST OF CHINA STATE SHIPBUILDING CORP)
- Filing Date
- 2026-02-12
- Publication Date
- 2026-06-02
AI Technical Summary
Ocean-going vessels face difficulties in effectively utilizing integrated air-space-ground networks for AI task computation offloading during voyages. Traditional methods struggle to quickly generate targeted network planning and resource allocation schemes, particularly exhibiting inefficiencies and instability issues during satellite switching and resource allocation.
By employing a multi-satellite assistance approach, suitable assisting satellites are selected based on environmental parameters and AI mission requirements. A computational offloading scheme is generated by combining multi-agent reinforcement learning algorithms, thereby achieving efficient utilization of satellite resources and precise mission matching. This includes optimization of satellite selection, switching, and access processes.
It improves the efficiency and stability of computational unloading, ensures the continuity and accuracy of task execution, reduces resource waste and transmission bottlenecks, and enhances the overall efficiency and reliability of AI task computational unloading for ocean-going vessels.
Smart Images

Figure CN122138220A_ABST