Dynamic data reconciliation algorithms and state evaluation methods for micro gas turbine operating parameters

Dynamic reconciliation algorithms and state evaluation methods for operating parameters are proposed to address measurement errors and uncertainties under variable operating conditions in micro gas turbine systems. Two improved dynamic data reconciliation algorithms are developed based on the Unscented Kalman Filter (UKF): the Savitzky-Golay Enhanced UKF (SGE-UKF) with sliding-window filtering and the Robust Locally Weighted UKF (RLW-UKF) with localized weighting. The results show that the improved algorithms are superior to the standard UKF. Under step change conditions, the SGE-UKF algorithm performs optimally, with a 67.08 % reduction in the root mean square error (RMSE) of the main parameter estimates. Under ramp change conditions, the RLW-UKF algorithm performs optimally, with a 76.93 % reduction in the RMSE of the main parameter estimates. Under sinusoidal change conditions, the estimated RMSE of the main parameters for the UKF, SGE-UKF and RLW-UKF algorithms decreased by 62.20 %, 71.23 %, and 71.51 %, respectively. Under all three operating conditions, the proposed algorithms demonstrate higher accuracy and data reliability compared with raw measurements. The results are applicable to condition monitoring of micro gas turbines, effectively improving the accuracy of parameter measurements and data reliability. This work provides robust data support for the real-time operation and experimental research of micro gas turbines.