Identification of Torsional Dynamic Parameters of Turbine Generator based on Hilbert Transformation

Abstract

An experimental method is proposed to identify torsional dynamic parameters from the measured impulse torsional response signals of turbine generator. Five-point cubic smoothing method is used to denoise the impact response signal. The envelope of the response signal is extracted by Hilbert transformation. The torsional natural frequency is calculated from the instantaneous phase signal. The envelope is fitted linearly in logarithmic coordinates. The torsional damping ratio and impact response amplitude are extracted. The torsional impulse response of a 600 MW supercritical turbine generator unit under three-phase short circuit fault is simulated. The torsional parameters are extracted by the method. The identification accuracy of damping ratio and natural frequency is high. It can be used to identify torsional parameters. The larger the amplitude of torsional response, the higher the identification accuracy. The measurement points should be arranged near the maximum deformation point of the modal to be identified.