The Research On Automatic Start-up Control Technology Of Reactor System

At present,operators mainly complete the start-up process of nuclear power plant reactors manually.The start-up process involves many systems,the steps are complex and time-consuming,and it is prone to human accidents,which seriously threatens the safety of nuclear power plants.In order to ensure the safety of the reactor start-up process,reduce the working pressure of operators,and promote the development of intelligent operation technology of nuclear power plants,it is necessary to carry out research on automatic start-up control technology.Sodium-cooled fast reactor is taken as the research object of this paper.The automatic start-up system of sodium-cooled fast reactor from the thermal shutdown state to the critical state is studied in this paper.A safety monitoring system is designed based on a deep neural network.The important safety parameters of the reactor system are used as network input.Through the calculation of the safety-monitoring network,real-time safety monitoring,evaluation and prediction of the reactor are carried out.Based on the deep Q-Learning algorithm,the rod position controller of the compensation rod group is designed.The deep reinforcement-learning algorithm is used to calculate the best height of each rod from the aspects of speed and safety,and according to the safety standards of the safety monitoring system Calculate the best waiting time after each lifting.After the algorithm was designed,the training algorithm of each controller was programmed and trained offline using JAVA language,and then rewritten into a FORTRAN program and a sodium-cooled fast reactor simulation model for joint adjustment tests,and the controller parameters were fine-tuned online.In order to verify the control effect of the automatic reactor start-up system,an automatic start-up simulation experiment was carried out under the real-time monitoring of the safety monitoring system.The analysis of control rod position,reactor cycle,reactor core reactivity,reactor neutron count rate and other parameters in the process of raising safety rod group,regulating rod group,and compensating rod group gradually reaching the critical state,verified the design of this paper The automatic start-up system can quickly and safely complete the automatic start-up task.By comparing with the expert manual start-up process,it is found that under the action of the deep Q-Learning rod position controller,the automatic start-up system process takes shorter time and has higher start-up efficiency.The automatic start-up system designed in this paper can realize the safe and efficient completion of the automatic start-up process of the sodium-cooled fast reactor simulation system.The control strategies and intelligent algorithms involved can provide a certain reference for the research of the automatic start-up of the reactor system.