化学工程2026，Vol.54Issue(1)：52-58,7.DOI:10.3969/j.issn.1005-9954.2026.01.009
物理神经网络下的高温球床辐射换热
Heat transfer in packed pebble beds based on physics-informed neural network
摘要
Abstract
In generation Ⅳ advanced nuclear reactors,the radiative heat transfer between fuel particles determines the temperature distribution of the pebble bed,which is closely related to the core safety.However,the determination of radiation related parameters,such as absorption coefficient and radiative diffusion coefficient,is full of challenges.In terms of theoretical modeling,the pebbel bed radiation model was derived from the particle scale,and the radiation absorption coefficient was determined by combining the first-order approximation of the radiative transfer equation,and a new thermal conductivity-radiation coupling model was established by combining the contact thermal resistance model.The effective prediction of temperature distribution was realized by automatic solution of physics-informed neural network,and the radiation diffusion coefficient of the pebble bed was obtained,while the influence of heating power on heat transfer was discussed,and the relationship between the packing density and the radiation diffusion coefficient was established.The results show that thermal radiation has an important effect on the temperature distribution of the pebble bed,and the thermal radiation diffusion coefficient of thepebblebed for the high-temperature experiments is about 2.35×10-2 m.In addition,the effect of thermal radiation decreases when the heating power decreases.The packing density is negatively correlated with the radiation diffusion coefficient,and this approach provides a new perspective for intelligent solutions to complex thermal problems in the future.关键词
物理神经网络/球床热辐射/吸收系数/辐射扩散系数/堆积密度Key words
physics-informed neural network/radiative heat transfer of pebble bed/absorption coefficient/radiation diffusion coefficient/packing density分类
能源科技引用本文复制引用
午波旭,吴浩,郭张鹏,牛风雷..物理神经网络下的高温球床辐射换热[J].化学工程,2026,54(1):52-58,7.基金项目
国家自然科学基金资助项目(12105101,12027813) （12105101,12027813）
