2026年“核理安邦”联合博士生学术论坛

一种在MARC级联下的Q迭代优化方法

Not scheduled

1h

地址：清华大学校内

海报展示 海报展示 海报展示

Speaker

承辉 庞

摘要

碳的同位素C-13在能源、医疗以及环境等领域拥有广阔应用前景。本研究以碳同位素的分离为目标，基于MARC模型级联提出了一种融合模糊控制算法的自适应松弛因子优化方法，用于提升Q迭代过程的收敛性能。通过构建高斯-三角混合隶属度函数，设计模糊控制器，并采用Mamdani最小-最大推理机制，实现了对级联丰度分布迭代过程的智能优化。该方法通过动态调节松弛因子，有效提升了迭代收敛效率。以二氧化碳作为分离对象，通过不同分离级数和收敛精度的对比验证了模型的有效性。结果表明，本研究设计的数值优化算法不仅具有较好的收敛加速效果，还展现出良好的鲁棒性，可以为Q迭代方法的高效求解提供新的优化策略，提高迭代效率。

Abstract

The carbon isotope C-13 has broad application prospects in the fields of energy, medicine and environment. This research aims at the separation of carbon isotopes and proposes an adaptive relaxation factor optimization method based on the MARC model cascade and integrating the fuzzy control algorithm, which is used to improve the convergence performance of the Q iteration process.By constructing a Gaussian-triangular hybrid membership function, designing a fuzzy controller, and employing the Mamdani minimum-maximum reasoning mechanism, the intelligent optimization of the iterative process of the cascaded abundance distribution was achieved.This method effectively improves the iterative convergence efficiency by dynamically adjusting the relaxation factor. Using carbon dioxide as the separation target, the validity of the model was verified through comparisons of different separation stages and convergence accuracies.The results show that the numerical optimization algorithm designed in this study not only has a good convergence acceleration effect, but also demonstrates excellent robustness. It can provide new optimization strategies for the efficient solution of the Q iteration method and improve the iteration efficiency.