Adaptive Refinement Simulation Model of Condenser Based on the Mechanism-AI Coupling Modeling Paradigm

Abstract Condenser is widely regarded as the key equipment of nuclear energy utilization device, and its performance has an important influence on the safety, economy and stability of the whole nuclear energy utilization device. In this paper, aiming at the problems that the simulation accuracy of the condenser real-time simulation model is lower than that of the numerical calculation simulation method and it is difficult to realize self-adjustment according to the actual changes, the NSGA-III ( a multi-objective optimization algorithm ) related to machine learning in AI technology is introduced into the mechanism modeling method,, that is, a mechanism-AI coupling modeling method is adopted. The specific model construction process is divided into three steps : constructing the basic model of the condenser according to the fluid network method, using the AI algorithm to determine the reason for the reduction of the model accuracy, and coupling the mechanism model with the AI algorithm to construct the condenser adaptive refinement simulation model. Finally, the simulation of a condenser is carried out to verify the model. The results show that the mechanism-AI coupling model can further improve the simulation accuracy of the condenser under the condition of meeting the real-time simulation requirements. When the condenser parameters change, the model can be adjusted independently to reduce the error caused by the change of the condenser parameters.