Research on nonlinear generalized predictive control method for reactor core power based on parameter identification

Reactor core power control technology is the core technology for load tracking in nuclear power plants. The problem of parameter degradation may occur during long-term reactor operation, and most predictive control methods rely on accurate reactor parameter models for prediction, so the control effect may be affected when the parameters are degraded, thus affecting the accuracy and stability of load tracking, and may even pose a potential threat to the safety of the system. To address this problem, this paper proposes a control strategy that integrates nonlinear generalized predictive control with parameter identification based on the Levenberg-Marquardt optimization algorithm (NGPC + Identification) to cope with the adverse effects of reactor model parameter degradation on control performance during long-term operation of nuclear power plants. In this paper, the reactor core nominal model and parameter degradation model are established based on neutron dynamics, thermo-hydraulics and reactivity mechanism equations, and then the degraded parameters of the reactor are obtained through parameter identification, which are acted on in the NGPC controller to realize the reactor load tracking. Finally, the control methods of MPC (Model Predictive Control), NGPC, and NGPC +Identification are compared by load tracking experiments. The results show that the NGPC + identification control strategy proposed in this paper performs better in terms of the load tracking effect.