Model predictive control for automatic operation of space nuclear reactors: Design, simulation, and performance evaluation

Space nuclear reactors possess the inherent benefits of independence from solar energy and the capability to withstand intricate external circumstances. Distinguishing themselves from terrestrial reactors, space nuclear reactors function within the realm of outer space, operating autonomously throughout their entire operational cycles. Consequently, the establishment of a dependable automatic control system assumes utmost significance. In this paper, a high-fidelity nonlinear space thermionic nuclear reactor model is established, encompassing neutron physics, thermal-hydraulic, and thermoelectric conversion system models. An automatic control system with a model predictive controller (MPC) is designed based on the system. A simplified state-space model is established based on system identification as the internal model for the MPC controller, demonstrating good agreement with the dynamic response of the nonlinear system model. The deployment of the MPC controller for system control simulations exhibits excellent control performance under various operating conditions. In the case of a nuclear power step response, the MPC controller achieves an overshoot of 0.07 % and a settling time of 0.65 s. For an electric power step response, the MPC controller yields an overshoot of 0.28 % and a settling time of 36.6 s. These control performances are significantly better than those achieved by the PID controller. In power tracking control, the MPC controller demonstrates virtually error-free tracking performance. When external reactivity disturbances are introduced, the MPC controller outperforms the PID controller in swiftly compensating for the disturbances, maintaining the system's operating state at the desired level.