Reactor lightweight shielding optimization method based on parallel embedded genetic particle-swarm hybrid algorithm

Abstract

Reactor lightweight shielding design is a multi-objective optimization problem, which must balance multi-dimensional design parameters such as dose, weight and volume. Combining the strong global search ability of the genetic algorithm (GA) and the fast convergence speed of the particle swarm optimization (PSO) algorithm, we proposed a Parallel Embedded Genetic Particle-swarm Hybrid Algorithm (PEGPHA) for reactor lightweight shielding optimization. After parallelizing GA and PSO generation by generation, reorganizing the population for each generation, and updating the key parameters in the iterative equations, the respective advantages of GA and PSO are balanced. Applied and tested on a small helium-xenon cooled reactor, PEGPHA has a higher convergence rate than GA and has a higher optimization efficiency than PSO. The number of ideal solutions obtained by PEGPHA is 1.23 times higher than GA, 2.22 times higher than PSO, and the average fitness of the ideal solutions is reduced by 2.9% and 2.8% compared with GA and PSO, respectively. The average optimization depth of PEGPHA is also shown to be 1.40 times and 2.10 times higher than that of GA and PSO. PEGPHA provides a larger ideal space for choosing high-quality solutions that are lighter in weight and smaller in volume while meeting dose limits, showing great application potential.