LAUSR

ABSTRACT This paper aims to propose a new methodology for optimizing fuel loading pattern in a nuclear reactor which is important for its higher safety and economic efficiency. Previous researches have proposed various methodologies to decide better loading patterns automatically. However, the processes still require manual operations of engineers to automatically design actual loading patterns. Swarm intelligence algorithm has currently gained interest as a solution to seek the patterns. Although these methodologies generate better patterns, they sometimes struggle with getting out from local optima and fails to complete the optimization. Large and multimodal solution space sometimes captures worse solutions due to local optima. The conventional methodologies struggle with setting proper parameters to get out from local optima. This research focuses on Multi-Swarm Moth Flame Optimization with Predator (MSMFO-P), an improved Moth Flame Optimization (MFO) by applying the concepts of predator and multi-swarm, as new methodologies. The method of MSMFO-P was applied to solve a loading pattern problem and compared with the conventional optimization methods such as simulated annealing (SA), Hybrid genetic algorithm (GA), and particle swarm optimization (PSO). The results of our experimental works indicated that MSMO-P generates better loading patterns than the conventional methodologies.