The Development Of New Optimization Algorithms And Applications In Optimal Design For Nuclear Power Plant

Nuclear power has been used in ship and electricity generation industry for its advantages of low conventional pollutants and greenhouse gas emission,air independence and high energy density.However,such disadvantages as relative large size,low thermal efficiency and high investment requirement can also be found in most popular nuclear power plant with pressurized water reactor due to the relatively low specific enthalpy of the fresh steam supplied to steam turbine as opposed to fossile fuel plant.On the premise of satisfying the performance requirements,the optimal design of nuclear power plant enables the decrease of weight,volume and investment and the increase of thermal efficiency,which can effectively lower the difficulties of manufacturing and transportation of components of nuclear power plant and improve the economics and competitiveness.Hence,it is of great significance both in theoretical and engineering aspects.Nevertheless,the design optimization of nuclear power plant can be ranged into complex nonlinear constrained optimization problem because of the complexity and intense coupling of the component and system.To realize the design optimization of either nuclear power plant or its components,proper optimization algorithm with good performance is indispensable.In this work,two novel optimization algorithms suitable for nuclear power plant optimization were developed.The mathematical models and the corresponding evaluation codes for both typical nuclear power component and whole system were established and developed.Based on the in-house developed algorithms and the models,case studies for optimization were carried out.The main contents and results are briefly introduced as follows:1.The design and development of novel hybrid genetic algorithm?NHGA?.To overcome the drawbacks of traditional genetic algorithm,one parallel crossover-mutation strategy and three adaptive genetic operators were proposed to improve the convergence performance and searching effiency.In addition,a cyclic reflection operator and a cyclic expansion operator were proposed to improve the local deep-search ability of Nelder-Mead simplex algorithm.In view of the complementary mechanism of genetic and Nelder-Mead simplex algorithm,a novel hybrid genetic algorithm was developed by properly combining the two modified algorithms.Furthermore,a normalized adaptive relaxation-constraint processing method was proposed in the new algorithm in order to improve the ability of dealing with constrained optimization problems.The NHGA was tested with 8 unconstrained benchmark problems,13 constrained benchmark problems and 4 engineering optimization problems.And the performance comparison with other state-of-the-art algorithms was also carried out.The effectiveness and feasibility of NHGA was verified.2.The development of novel hybrid genetic algorithm for multi-objective optimization?NHGA-MO?.A fast?-dominated sorting strategy and a dynamic crowding distance strategy were proposed to improve the distribution performance of optimized solutions.In addition,the Nelder-Mead simplex algorithm was adaptively modified to improve the convergence of solutions.Integrating the modified strategies into the NHGA algorithm,a novel hybrid genetic algorithm designed for multi-objective optimization problems was developed.The NHGA-MO was tested with 7 unconstrained benchmark problems,6 constrained benchmark problems and4 engineering optimization problems.And the performance comparison with other state-of-the-art algorithms was also carried out to verify the effectiveness of the proposed algorithm.3.The case studies were conducted for single-objective and multi-objective optimization of typical component of nuclear power plant.The feedwater heater was selected as a typical component for optimization design.The mathematical model of feedwater heater in nuclear power plant was established and the corresponding evaluation code was developed using C#language.The model and the code were validated by using published data.Based on the model and the new algorithms,the single-objective and multi-objective optimization of weight,volume and total cost of the 7^th high-pressure and 3^th low-pressure feedwater heater in Daya Bay nuclear power plant were implemented respectively with reasonable performance constraints.The results of single-objective optimization show that the weight,volume and total cost of the feedwater heaters in optimized scheme can be reduced by 13.46%,19.88%,21.45%and 15.29%,27.79%,22.03%,respectively.The multi-objective optimization result offers a series of non-dominated optimized schemes of the feedwater heaters,which can be selected according to the design requirements.4.The single-objective and multi-objective optimization of thermodynamic system of nuclear power plant was also performed.The mathematical model of thermal system in nuclear power plant was established and the corresponding evaluation code was developed.The model was validated against the data for Daya Bay nuclear power plant.Based on the developed model and the new algorithms,the single-objective and multi-objective optimization of the thermal system in Daya Bay nuclear power plant was carried out given some thermal-hydraulic constraints.For the single-objective optimization,only the operation parameters were optimized to maximize the power output and the configuration and structure of components were kept unchanged.Compared with the original design,the power output of the optimized scheme increases by 10.3 MW,the corresponding thermal efficiency increases from 33.81%to34.16%.For the multi-objective optimization,both the system operation parameters and the feedwater heater structure parameters were optimized.The results show that the optimized scheme can yield the increase in power output as much as 23.9 MW,the corresponding thermal efficiency increases from 33.81%to 34.63%.