Research On Model Predictive Control For The Reactor Power Control System Of A Nuclear Power Plant

Modern nuclear power plant must have the ability to response to the power demand on the grid rapidly, however, generally, the nuclear power plants are complicated and nonlinear, and the parameters of the system may vary with the operation point, hence,the system is uncertain. Additionly, on account of the technical restrictions and the operation security, we must attach importance to the constraints existing in the system.Therefore, it is a great challenge for a reactor power control system, especially when wide-range power variations occur in the load following condition.In this paper, the Three Mile Island (TMI)-type pressurized water reactor (PWR) model is selected as the research model, T-S fuzzy modeling on it is utilized to approximate the nonlinear plant by local linear models. Based on which a nonlinear fuzzy MPC controller is devised via parallel distributed compensation (PDC) scheme in order to accomplish the wide-range load following, and solve the constraint optimization problem effectively.But the T-S fuzzy model is uncertain with its varying parameters, in terms of this problem, a fuzzy robust model predictive controller is designed based on the uncertainty description by polytopes. Partial-offline fuzzy robust model predictive controller is designed by constructing a fuzzy Lyapunov function and making the current control as a free variable which is separated with the future control governed by state feedback control law, besides, polyhedral invariant set is utilized to expand the feasible region for the control law,Finally, satisfactory performance is obtained in the simulation for the reactor power control.