Fault-tolerant control for water-feeding of steam generator system with faults and uncertainties

Steam generator water level regulation in the pressurized water reactor (PWR) is a safety-critical task. Conventional model-based controllers deliver strong performance when the nominal dynamics are accurate, but their effectiveness can deteriorate quickly when feedwater valves develop leakage and when thermohydraulic characteristics vary with aging and operating point. To address these challenges, this paper develops a data-driven fault-tolerant control (FTC) framework for the steam generator feedwater loop. A physically structured feedback scheme is adopted and augmented with an online gain-update mechanism, thereby avoiding the need for the precise knowledge of the system model. An iterative data-driven procedure is formulated to compute stabilizing state-feedback gains directly from fault-free state and input trajectories, where the gain update is obtained by solving data-induced Lyapunov equations using least-squares estimation. During faulty or uncertain operation, auxiliary adaptive terms compensate bounded disturbances and actuator faults, and a uniform ultimate boundedness result is established for the closed-loop solutions. The approach is evaluated on a nonlinear simulation model of the secondary loop that includes a steam generator, feedwater valve, main steam header, pump, and turbine dynamics. Comparative studies under feedwater valve leakage faults, steam generator heat transfer degradation, and their concurrent occurrence demonstrate that the proposed controller preserves water level regulation with smaller deviations and faster recovery than a baseline PID strategy, thereby improving robustness and fault tolerance without relying on a precise plant model.