Load-following operation in nuclear power plants presents a significant control challenge because of reactor nonlinearities, time-varying parameters, disturbances, and the coupled neutronic and thermal-hydraulic dynamics, particularly those associated with axial xenon oscillations. These effects can degrade power-tracking performance and increase control-rod actuation, motivating the development of robust and adaptive control strategies for flexible regulation of nuclear power plant operation. In this study, a T-S fuzzy PI-based adaptive sliding mode composite controller is proposed to achieve smooth and precise power-level control of nuclear reactors. The designed controller combines a T-S fuzzy PI controller for nominal reference tracking with an adaptive sliding mode controller (SMC) for disturbance rejection and robustness against model uncertainties. The fuzzy mechanism continuously adjusts the PI gains and the SMC switching gain according to the power-tracking error and its rate of change, thereby adapting the control action to different operating conditions and reducing excessive discontinuous switching. The designed controller gains are optimized using an SQP strategy. A validated nonlinear two-point reactor kinetics model with three groups of delayed neutron precursors is utilized, incorporating normalized neutronics, thermal-hydraulics, reactivity feedback, and xenon-iodine poison dynamics for the upper and lower reactor regions. To validate the proposed controller, several simulations are performed under load-following transients, reactivity disturbances, and ±15% parameter uncertainties in significant reactor model parameters. The results of numerical simulations are compared with those from conventional sliding mode control and classical PI control with fixed parameters. The designed composite controller not only reduced the ITAE by 35.6% and the settling time by 28.4%, while decreasing control-signal chattering by 42.1% compared with the conventional sliding mode controller, but also improved reference tracking performance, faster settling, enhanced disturbance rejection, and reduced control-action oscillations, confirming the controller's adaptability and robustness under varying reactor operating conditions.

VVER-1000 nuclear reactor, Two-point reactor kinetics model, Load-following operation, Power-level control, Adaptive sliding mode control, Disturbance rejection.