A control-oriented reactor core model must preserve axial power redistribution while remaining computationally tractable for advanced controller design. This study develops a variable-node fractional-order axial reactor model and a parameterized dynamic mode decomposition with control (pDMDc) reconstruction framework for pressurized water reactor power-control applications. The full-order model combines axial power-distribution coefficients, thermal feedback, iodine-xenon dynamics, and distributed control-rod reactivity; node-number assessment identifies 20 nodes as an effective accuracy-cost compromise. Local DMDc operators are blended by radial-basis-function weights. Power-rod position schedules rod-induced nonlinearity, while coolant inlet-temperature perturbation is added as a second variable to distinguish thermal-feedback regimes at the same rod position. Across rod-motion and multi-condition tests, pDMDc reproduced total power, axial offset, nodal power, fuel temperature, and coolant temperature while reducing prediction time from hundreds or thousands of seconds to less than one second. In the combined rod-motion and inlet-temperature test, the selected 2 × 3 structure reduced power and coolant-temperature RMSEs by 48.4% and 46.8%, respectively, relative to the single-parameter baseline, while preserving the principal axial dynamics. The resulting model provides an efficient basis for MPC, optimal control, and learning-based reactor power-control studies.
 

Variable-node fractional-order reactor model; Nuclear power control; Reduced-order modeling; Parameterized DMDc; RBF scheduling