Background: In recent years, genetic algorithms have been applied in the field of nuclear technology design, producing superior optimization results compared to traditional methods. They can be employed in the design and optimization of beam shaping assemblies (BSA) BSA to obtain the desired neutron beams. But it should be noted that the direct combination of Monte Carlo methods with genetic algorithms requires a significant amount of computational resources and time.
Purpose: Design and optimize BSA more efficiently to achieve neutron beams that meet specified recommendations.
Methods: We propose an approach of NSGA II with crucial variables which are identified by multivariate statistical techniques. This approach significantly reduces the problem sizes, thus reducing the time required for optimization. We illustrate this methodology using the example of BSA design for AB-BNCT.
Results: The computational efficiency has tripled with crucial variables. By using NSGA II, we obtained optimized models conforming to both the new and old version IAEA BNCT guidelines through a single optimization process and subjected them to phantom analysis. The results demonstrate that models obtained through this method can meet the IAEA recommendations with deep advantage depth (AD) and high absorbed ratio (AR).
Conclusion: The genetic algorithm with crucial variables displays tremendous potential in addressing BSA optimization challenges.
Keywords: AB‐BNCT; Monte Carlo; NSGA II; beam shaping assembly; genetic algorithm.
© 2024 American Association of Physicists in Medicine.