Comparative Analysis of Genetic Algorithms Based on Different Selection Strategies for Refueling Optimization in the Ratio Method

Abstract

The genetic algorithm is one of the classic algorithms applied to the refueling optimization. An important part of this algorithm is the selection strategies. The existing studies often directly adopt the roulette wheel selection and stochastic tournament selection, and are lacking in comparison and analysis of different selection strategies. To obtain the selection strategy with the strongest optimization capability, this study, with the 1/6 core of a thorium-based prismatic high-temperature gas-cooled reactor (HTGR) taken as an example, constructs the fitness function in the ratio method, performs core physics calculation using the DRAGON code, and in conjunction with the elitism strategy, compares the optimization capabilities of the five selection strategies, including the roulette wheel selection, stochastic tournament selection, uniform ranking method, exponential ranking selection and deterministic selection. The study results show that the optimization capability of the exponential ranking selection is superior to the other four strategies, so the exponential ranking selection is most suitable for solving the refueling optimization problems.