| In recent years,electronic accelerator technology plays an important role in the fields of industry,agriculture and medical treatment.Besides,it has developed rapidly and created enormous economic and social benefits.The radiation shielding safety design is a key part in the process of accelerator design.It is a complex engineering optimization problem to find a shielding scheme that meets the requirements of small size,light weight,low price and friendly environment under the condition of ensuring that the dose meets the national regulations for the safe and reasonable design of accelerator radiation shielding.Traditional radiation shielding design schemes are usually carried out based on manual iteration and the experience of the designer,so some defects will be showed up,such as increased uncertainty,long design cycle,resulting in the radiation shielding design is not optimal,this design method can no longer meet the requirements of accurate accelerator design.Therefore,using modern multi-objective optimization theory and method to solve the problem of shielding system optimization design has become an urgent need for accurate shielding design.The intelligent optimization algorithm of multi-dimensional,multi-parameter and multi-objective shielding design is very important for the radiation protection design of accelerator and the safety of workers,as well as the development of industrial electron accelerator.Taking the non-destructive testing accelerator as the research object,this study uses the engineering CAD model of the accelerator and Monte Carlo software to carry out accurate modeling and simulation calculation of the accelerator,which serves as the basis for the multi-objective optimization design.Then the GA-BP neural network and the NSGA-? multi-objective genetic algorithm were combined,to established accelerator shielding design of multi-objective optimization method.Finally,by using MATLAB program algorithm toolbox and its programming ability,the optimal combination of shielding design parameters is obtained through global optimization within the given range of accelerator shielding design parameters.The main research contents and innovations are as follows:(1)Based on the engineering CAD model of nondestructive testing accelerator,Super MC and MCNP software were used to convert the engineering CAD model into a complete MCNP file after a series of preprocessing such as repair,simplification and transformation.Then,the distribution of radiation field around the accelerator is qualitatively analyzed by simulation calculation,and the dosimetric indexes of the protection performance of the accelerator are compared in accordance with relevant national standards.Based on this,the shielding design of the accelerator is optimized.(2)By using the genetic algorithm for neural network connection weights and threshold optimization,with GA-BP neural network method is established for accelerator shot blocking leakage rate is high efficient and accurate prediction model.Regard neural network prediction model as the fitness function of genetic algorithm optimization,successfully solved the speed of genetic algorithm and random optimization problem,greatly accelerated the accelerator shielding design optimization process.(3)Using non dominated sorting genetic algorithm with elitist strategy NSGA-? accelerator shielding design optimization strategy for case study,through the coordination of multiple design collaborative optimization of the relationship between the target and the make every design target is in a relatively optimal accelerator shielding design collection.Then,it is compared with the simulation results of Monte Carlo simulation software to verify the accuracy and effectiveness of the optimization algorithm.Combining the multi-objective optimization method of accelerator shield design with the experience of the designer,the multi-objective optimization design of accelerator shield can be realized more efficiently. |
PDF Full Text Request