Application of a hybrid method based on the combination of genetic algorithm and Hopfield neural network for burnable poison placement

In the last decades genetic algorithm (GA) and Hopfield Neural Network (HNN) have attracted considerable attention for the solution of optimization problems. In this paper, a hybrid optimization method based on the combination of the GA and HNN is introduced and applied to the burnable poison placement (BPP) problem to increase the quality of the results. BPP in a nuclear reactor core is a combinatorial and complicated problem. Arrangement and the worth of the burnable poisons (BPs) has an impressive effect on the main control parameters of a nuclear reactor. Improper design and arrangement of the BPs can be dangerous with respect to the nuclear reactor safety. In this paper, increasing BP worth along with minimizing the radial power peaking are considered as objective functions. Three optimization algorithms, genetic algorithm, Hopfield neural network optimization and a hybrid optimization method, are applied to the BPP problem and their efficiencies are compared. The hybrid optimization method gives better result in finding a better BP arrangement.

► The performance of GA, HNN and combination of them in BPP optimization in PWR core are adequate.
► It seems HNN + GA arrives to better final parameter value in comparison with the two other methods.
► The computation time for HNN + GA is higher than GA and HNN. Thus a trade-off is necessary.