Research On The Intelligent Radiation Shielding Design Method Using The Genetic Algorithm

The design of radiation shielding system is an important part of the nuclear engineering design, that is directly related to the radiation safety of nuclear installations and personnel, and has a great impact on the overall performance of the device and project cost, or even restricts the realization of the ultimate goal. Radiation shielding design for the modern state-of-the-art nuclear devices, such as: commercial nuclear power plant, complex nuclear facilities, large-scale accelerator and target station, especially for ship nuclear power reactors, space reactors, not only the radiation dose of specified area to meet the design limit, but also the project cost is acceptable. Even more, indicators of weight and size are stringent controlled which make the design process very complex and time-consuming. In addition, the selection of the shielding design is also affected by the knowledge of shielding designer and his engineering experience. Complex nuclear device radiation shielding design is actually a multi-dimensional, multi-objective optimization problem.The mature mult-objective radiation shielding optimization design theory for complex nuclear device radiation shielding design has not been established at home and abroad.In this thesis, a multi-dimensional and multi-target nuclear radiation shielding design optimization theory based on genetic algorithm (GA) has been established by combination of GA and the advanced shielding calculation programs. Based on the theoretical and in the light of the development process of software engineering, the multi-dimensional, multi-target smart radiation shielding design and optimization software platform has been developed to achieve the automatic selection of the radiation shielding design, eliminate the impact of human factors error. Then, intelligent radiation shielding optimization platform was tested strictly and applied to the CSNS target station main shielding and neutron channels shielding design, and also used in the shielding design of the secondary shield area for the Sefaner ship reactor. At last, the correctness and feasibility of the radiation shielding design theory was verified. The main research include:(1) Establish a multi-objective radiation shielding optimization design theory model based on GA. In this theory, hybrid coding has been used to express variables of radiation shielding optimization process to make the GA variable expression simple, variety of crossover and mutation model has been used to achieve genetic opearator. In order to eliminate the influence of different sub-objective functions on fitness function, the adaptive weight coefficients has been adopted to make the weight coefficient between the various sub-goals adjustable. The method for solving the various sub-goals are shown, such as:size, weight, cost, and dose rate. As one of the evaluation standards, the temperature limits is used for the optimization search process that ensuring the results of the optimization program realistically.(2) The establishment of a multi-dimensional fast speed coupling dose rate calculation method and succeeded in solving the problem of speed matters for GA optimization process. In the early iterations of the genetic population, the one-dimensional discrete ordinate method or3-D point kernel integral method is used for shielding calculations to obtain a preliminary optimization population quickly, then the high precision and time-consuming multi-dimensional discrete ordinates method or MC method is used for the final optimization model calculation, this approach can meet the both requirements of optimization process fast and the accuracy of optimization results.(3) The set range of genetic operators and population size parameters for fast radiation shielding design optimization have deen reserched. The setting range of this parameters offers an access to the engineering applications for the platform with provide experience.(4) The radiation shielded multi-objective optimization platform has been tested by the design examples, and NUREG/CR-6453pressure vessel experiments and NUREG/CR-6115(BNL-NUREG-52395) benchmarks have been used to test the multi-dimensional calculation module of the platform. At last, the theory was used for the shielding engineering design of large and complex nuclear device, such as:China Spallation Neutron Source target station shielding design and the ship reactor main shielding.The theory based on genetic algorithm for multi-dimensional, multi-target nuclear radiation shielding design has been developed and a software platform was developed based on it. The multi-target radiation shielding intelligent design theory and its software system correctness and feasibility of engineering applications has been verifed. Shielding design method based on genetic algorithm may partially replace the existing shielding design methods, thus opening up new avenues for radiation shielding design.