Models of heavy-duty gas turbines (HDGTs) established solely based on mechanisms or data cannot account for performance degradation over time

leading to increasing simulation errors. To achieve more accurate dynamic simulation results

data assimilation (DA) is applied to the HDGT dynamic simulation process. An online dynamic data-driven simulation system for HDGT is constructed based on DA technology. The unscented Kalman filter (UKF) algorithm is employed to online estimate four key parameters (including compressor outlet temperature

compressor outlet air pressure

turbine outlet gas temperature and gas turbine power) within the simulation model. Simulation results demonstrate that the UKF can utilize observed data to update the simulation model parameters in real time

allowing the model to track the actual system states effectively. Under simultaneous gradual variations (within 2%) of the compressor efficiency and flow characteristic coefficients

the maximum tracking error remains below 0.7%. For simultaneous step changes exceeding 2% in these coefficients

the tracking error exhibits rapid convergence. The simulation results demonstrate that this method employs online parameter identification to enable dynamic tracking of actual system performance by the simulation model

providing a feasible technical path for predictive control

real-time optimization

and digital twin applications of gas turbines.