Multi-layer perception based model predictive control for the thermal power of nuclear superheated-steam supply systems

Nuclear superheated-steam supply systems (Su-NSSS) produces superheated steam flow for electricity generation or process heat. Although the current Su-NSSS control law can guarantee satisfactory closed-loop stability, which regulates the neutron flux, primary coolant temperature and live steam temperature by adjusting the control rod speed as well as primary and secondary flowrates, however, the control performance needs to be further optimized. Motivated by the necessity of optimizing the thermal power response, a novel multi-layer perception (MLP) based model predictive control (MPC) is proposed in this paper, which is constituted by a MLP-based prediction model and the control input designed along the direction opposite to the gradient of a given performance index. It is proved theoretically that this MLP-based MPC guarantees globally-bounded closed-loop stability. Finally, this newly-built MLP-based MPC is applied to the thermal power control of a Su-NSSS, whose implementation is given by forming a cascaded feedback control loop with the currently existing Su-NSSS power-level control in the inner loop for stabilization and with this new MPC in the outer loop for optimization. Numerical simulation results verify the correctness of theoretical result, and show the satisfactory improvement in optimizing the thermal power response.