【摘要】 随着人工智能（artificial intelligence,AI）的飞速发展，其在核燃料及材料领域的应用正逐渐成为推动核能科技进步的新动力。本文全面回顾了AI在核燃料及材料领域的研究现状，并对未来的发展趋势进行了深入分析。首先介绍了应用于科学研究的AI方法，分别从网络架构和学习范式2个方面展开论述。其次系统总结了AI在材料级和整体级的性能预测、材料和结构的设计优化和燃料生产运行过程的视觉任务3个方面的应用现状。随后展望了AI与核燃料及材料结合的未来发展趋势，在算法层面，讨论了提升机器学习模型的可解释性、量化不确定性的方法，以及有限监督学习技术在减少数据需求方面的重要性；在应用层面，讨论了多尺度多物理场仿真加速、拓扑优化与生成式设计、核材料性质通用预训练模型，以及自动化实验室等关键技术。最后为进一步促进AI在核燃料及材料领域的应用提出了几点建议。更多还原

【Abstract】 With the rapid development of artificial intelligence(AI), its application in the field of nuclear fuel and materials is gradually becoming a new driving force for the advancement of nuclear energy technology. This article comprehensively reviews the current state of AI research in the field of nuclear fuel and materials and conducts an in-depth analysis of future development trends. It first introduces AI methods applied to scientific research, discussing from two aspects: network architecture and learning paradigms. Next it systematically summarizes the current state of AI applications in performance prediction at the material and structure levels of nuclear fuel materials, design optimization of materials and structures, and computer vision in fuel production and operation. The review then looks forward to the future development trends of the combination of AI with nuclear fuel and materials. At the algorithm level, it discusses methods to enhance the interpretability of machine learning models, quantify uncertainty, and the importance of limited supervised learning technology in reducing data requirements. At the application level,it discusses key technologies such as acceleration of multi-scale multi-physics field simulations, topological optimization and generative design,universal pre-trained models for nuclear material properties, and automated laboratories. Finally, several suggestions are proposed to further promote the application of AI in the field of nuclear fuel and materials.更多还原