Research On Reliability-based Design Optimization Method Of Passive Safety System

Limited by space size and load capacity,the design of safety systems for the marine nuclear power plants needs to be economical and reliability.As a new generation of inherent safety technology,passive safety system can take into account the safety and economy of nuclear power units,and its wide application has become an inevitable trend in the development of the marine nuclear power plants.Because passive safety systems are driven by physical laws such as natural cycles,they are vulnerable to uncertainty.In order to avoid the risk caused by uncertainty,it is necessary to carry out research on reliability design optimization methods considering the influence of uncertainty of passive safety system design parameters.Based on the sequence optimization and reliability assessment method(SORA),a reliability-based design optimization method for passive safety systems has been proposed in this paper.This method performs design optimization and reliability analysis sequentially.In the design optimization part,the optimization algorithm is used to calculate the optimization and obtain the optimal solution of the objective function.For the reliability analysis part,the inverse reliability analysis is carried out at the optimal solution of the objective function to solve the inverse most probable point of the passive safety system.By the inverse most probable point,the shifting vector is solved to adjust the constraint function in the design optimization process,and realize the coupling of reliability analysis and design optimization.Aiming at the characteristics of many design parameters,high nonlinearity and multiple constraints in the design optimization of passive safety systems,this paper proposes a two-layer adaptive genetic algorithm based on adaptive penalty function.The calculation of the test function of the constraint optimization problem shows that compared with the genetic algorithm based on the penalty function,the two-layer adaptive genetic algorithm based on the adaptive penalty function has more advantages in solving the constraint optimization problem,the error is smaller,and the accuracy is higher in the design optimization of passive safety systems.Aiming at the problem that the computational cost is too high due to the need to call the thermal hydraulic analysis program multiple times during the design optimization process,this paper uses a two-layer adaptive genetic algorithm to optimize the initial weights and thresholds of the BP neural network based on the BP neural network model,and constructs a TAGA-BP high-precision surrogate model.Aiming at the shortcomings of the difficulty of global accurate modeling in the training process of TAGA-BP model,an optimally solution-driven local TAGA-BP model is developed,which is used instead of the thermal hydraulic program to calculate the constraint function response value to optimize the reliability design of passive safety system.The design optimization method proposed in this paper is applied to the reliability-based design optimization of passive residual heat removal system of IP200 marine nuclear power plant.Under the limited probability of failure,the weight of the passive residual heat removal system is reduced by2.458×10~3 kg compared to the initial value,which is reduced by 3.44%.The method proposed in this paper takes into account the influence of design variable uncertainty when carrying out design optimization,has the advantage of fast calculation speed,serves the design optimization of passive safety system,can reduce the weight of passive safety system,and is of great significance for improving the economy of passive safety system.