| Irradiation processing technology is a non-powered nuclear technology that utilizes radiation energy to treat the goods in order to achieve desired effects.In a gamma irradiator,the dose field distribution is determined by the loading pattern of the radioactive source pencils.Reasonable optimization of the loading plan can help to obtain a uniformly distributed dose field with high energy utilization efficiency,which will improve the processing quality and efficiency and boost the economic benefits of the irradiator.However,the optimization of the loading plan is a typical combinatorial optimization problem with non-convex,discontinuous and multi-modal solution space,and the number of feasible solutions tends to grow exponentially with the amount of installed source pencils.There is currently no universal solution within the irradiation industry and most irradiators still rely on manual experience for source planning that cannot cope with the complex demands of large commercial gamma irradiators.In this thesis,a genetic algorithm-based optimization method for the source loading plans is proposed.By analyzing and modelling the source loading problem,a specific implementation procedure of this optimization method is described.In the optimization process,the loading plan is considered as an individual in the population,and the value of the objective function to be optimized(e.g.dose uniformity ratio or average dose rate)can be mapped to the fitness value of the individual.Then based on fitness values,the genetic algorithm evolves individuals with high fitness and good performance,generation by generation,to complete the optimization.Simple genetic algorithm(SGA)can be used for the single-objective optimization of the loading plans.The validation result of a practical optimization example shows that the loading plan optimized by SGA can be improved by up to 16.9% in terms of dose uniformity ratio.For multi-objective optimization,this thesis proposes and validates an approach using a fast non-dominated sorting genetic algorithm(NSGA-Ⅱ),which gives a set of Pareto optimal solutions and allows the decision maker to choose according to his preference for different optimization objectives.The optimization result for an study case shows that the optimized solutions gained by NSGA-Ⅱ resulted in a 16.0% improvement in dose uniformity ratio and 5.4% enhancement in average absorbed dose compared to the most probable random plan.The two study cases fully demonstrate the feasibility and effectiveness of the genetic algorithm-based optimization method proposed in this thesis.In addition to the algorithm development,this thesis also implements a Web program for dose field optimization of gamma irradiators.The program integrates two core functions,genetic algorithm-based optimization and Monte Carlo method-based simulation for the source loading plans,and provides a one-stop solution for source loading plans optimization problem. |
PDF Full Text Request