Research On Steam-induced Vibration Fault Of Steam Turbine Generator Set Early Warning Method

Based on a large number of domestic and foreign turbine steam-induced vibration fault cases, the paper deeply study the steam-induced vibration fault of turbo generator set. Studying the fault mechanism of steam-flow exciting vibration. Combining the fault cases, analyzed the fault typical feature, used the FMEA and FTA methods to study the steam-induced vibration fault, Through studying some steam-flow exciting vibration fault cases, fault causes and measures are summarized. Finally the steam-induced vibration fault early warning is proposed. Through vibration signal extraction, We obtained the fault signal characteristics. Based on the power plant’s practical vibration data, this paper proposed the vibration prediction methods, for instance Autoregressive and Moving Average(ARMA) model and second-order Volterra adaptive filter, firstly using the two models to predict vibration time series respectively. Then integrated the different methods and make a comparison of the two prediction results.Before using the model ARMA, we must ensure the vibration data is reposeful, this requires judging the reposeful of the time sequence. Through difference method, getting the reposeful data, then define the orders and parameters of the ARMA model, then verified the adaptability of the obtained ARMA model. Based on the AR、MA、ARMA three different models, this paper use the "fitting degree", "correlation coefficient" and "statistical analysis of the data of deviations" three standards to compare the superiority and inferiority of the model. In the second part, based on the second-order Volterra adaptive filter, firstly using the principle of phase space reconstruction, determined the delay time and embedded dimension, analyzed and compared the prediction effects of the different reconstruction parameters; finally, based on the same time sequence, this paper make a comparison of ARMA model and Volterra model.Through the comparison of the two methods, then combined the vibration prediction and steam-induced vibration fault, proposed the steam-induced vibration prediction procedure.