基于遗传算法的可燃毒物设计优化方法研究

可燃毒物可补偿寿期初过剩反应性及展平功率分布，因此对堆芯燃料组件设计具有重要意义。目前传统的优化设计主要依靠设计者的主观经验及判断，复杂耗时，其设计效率及可靠性急待改进。本文将多目标并行遗传算法应用于压水堆组件毒物选型优化，以反应性控制、功率分布和不同时期燃耗剩余等为目标，对可燃毒物材料类型、含可燃毒物燃料棒排列方式、毒物含量、轴向分层等决策变量进行优化，研究了遗传算法在燃料组件毒物多目标优化设计中的理论模型及实现方法。同时将遗传算法与蒙特卡罗粒子输运方法有机结合，应用到压水堆燃料组件设计中，得到了组件可燃毒物优化方案。针对二维和三维燃耗计算，分别筛选了13和40种优化方案。计算结果表明：Er2O3用作毒物的综合效果最好；Gd2O3、Eu2O3和Sm2O3的应用需结合堆芯方案开展进一步研究；HfO2和Dy2O3不适合用作可燃毒物。该结果与通过人工搜索优化得到的结论基本一致。同时，三维轴向分层可为优化提供更多可选的材料种类方案，以部分毒物的分层布置方式可减小功率峰因子。本文研究为堆芯燃料/毒物设计提供了先进方法及工具。

Optimization Method of Burnable Poison Design Based on Genetic Algorithm

The design of burnable poisons (BPs) can compensate for excess reactivity at the beginning of lifetime of nuclear reactors and flatten power distribution, which is especially important for nuclear reactors assembly design. At present, the traditional optimization design mainly relies on the subjective experience and judgment of designers, which is complicated and time consuming. Therefore, the efficiency and reliability of BPs design urgently need to be improved. In this paper, the multi objective parallel genetic algorithm (GA) was applied to the selection and optimization of the components of pressurized water reactor (PWR). Taking the reactivity control, power distribution and poisons residues in different periods as the objectives, the decision variables such as the type of BPs materials, the layout of fuel rods containing BPs materials, the purity of BPs and the axial division were optimized. Then optimization program was developed by combining multi objective parallel GA with Monte Carlo particle transport code RMC as the neutronics and depletion solver. 13 and 40 optimization schemes were selected for two and three dimensional burnup calculation. The results show that the comprehensive effect of Er2O3 as BP is the best. The application of Gd2O3, Eu2O3 and Sm2O3 should be further studied in combination with the core scheme. HfO2 and Dy2O3 are not suitable for using as BPs. The results are basically consistent with the results obtained by manual search optimization. At the same time, three dimensional axial division can provide more alternative material types for optimization, and power peaking factor can be reduced by layering some BPs. This paper provides useful methods and tools for BPs design of nuclear reactors.