Research On The Cyber-physical System Of Nuclear Reactor Power Based On Varying Load Condition

In the process of China's energy structure adjustment,nuclear power has played an important role in meeting national energy supply security and implementing energy supply-side structural reforms.With the advent of the industrial information age and the increasing installed capacity of nuclear power,it has great research and practical value that how to combine the fast-developing information technology with the traditional control of nuclear power plants,in order to meet the needs of nuclear power plants participating in power grid frequency modulation under the premise of ensuring the safe operation of nuclear power plants.Aiming at the non-linearity of the reactor power model and the mechanical constraint of the control rods during the variable power operation of the nuclear reactor,this paper applies the technology of Cyber-Physical to control the nuclear reactor power,to meet the purpose of enabling the reactor power to track the set value quickly and accurately under the condition of ensuring the safe operation of the nuclear reactor.First,the nuclear reactor power model was established based on the neutron dynamics and the reactor thermal hydraulics,and the reactor power data at different power levels was obtained based on the nuclear power simulator of Fuqing M310-1000 WM,then the least squares algorithm was used to identify the reactor power model parameters at different power levels.Secondly,aiming at the problem of model non-linearity and control rod constraint during variable power operation of the nuclearreactor,design a generalized predictive controller with the capability of model parameter online identification,and update the parameters at each control moment based on the input and output data of the system.In order to solve the problem that the traditional generalized predictive controller lacks the ability to deal with the constraints of control value effectively,a particle swarm algorithm improved by sinusoidal chaos strategy and adaptive inertia weight is used to implement rolling optimization of generalized predictive control,so that the controller has the ability to handle the constrain of control rod.Then,in order to ensure the safety of the nuclear reactor during the operation of variable power,a nonlinear support vector machine is designed for fault diagnosis of the nuclear plant coolant system.Based on the nuclear power simulator,160 samples were obtained under the two faults of the small break of the SG heat transfer tube and the small break of the loop hot section tube,and used to train and test the diagnostic performance of the nonlinear support vector machine.Finally,building the Cyber-Physical System of the nuclear reactor.An improved generalized predictive controller with constraint processing capability is set as the physical layer,which is used to achieve that the reactor power can quickly and accurately follow the set value during variable power operation and improve the response speed of the nuclear reactor.The multi-class non-linear support vector machine set as information layer,during the nuclear reactor operation,the four parameters of the regulator water level,regulator pressure,containment temperature,containment pressure are monitored to ensure the safety of the system.The MATLAB platform was connected to the Fuqing M310-1000 WM nuclear power simulator through the OPC protocol interface to build a simulation platform for the Cyber-Physical System of core power under the condition of variable power.