Intelligent Edge-Driven Rainwater Radioactive Purification System: Dynamic Optimization Design Based on Self-Supervised Learning

With the increase in global nuclear energy utilization, the removal of radionuclides in rainwater is becoming increasingly important for ensuring water safety, but existing nuclear wastewater treatment facilities are inefficient in this regard. This study designed a new rainwater radioactive element removal system, which adopts a large-span air membrane structure, has mobility and rainwater collection functions, and can be used for rainwater storage, transportation and radionuclide purification. The system uses porous materials as the main purification method and introduces a self-supervised learning method. Through a custom loss function, the purification parameters are automatically adjusted according to real-time monitoring data to improve the nuclide removal efficiency. Membrane separation technology in cycle method is a backup solution for specific nuclides that cannot be removed by porous materials. Concurrently, the system integrates renewable energy and smart material technologies, providing an economically viable, efficient and flexible nuclear pollution control program for Pacific coastal countries. Self-supervised learning uses the system’s own data to generate supervision signals, and the custom loss function is optimized for the task. Compared with the traditional fixed parameter purification system, this system can adapt to the changes in radionuclide concentrations in different environments, and through interdisciplinary integration, it improves purification efficiency and reduces costs, bringing new ideas to nuclide removal technology and demonstrating the potential of interdisciplinary technology integration to ensure water safety.