Long-Term Vibration Trend Prediction of Rotor System State Based on Support Vector Regression and Discrete Wavelet Decomposition

Abstract

In this paper, an new method is proposed based on support vector regression (SVR) and discrete wavelet decomposition (DWD) for long-term rotor vibration trend forecasting. The feasibility of SVR in long-term vibration trend forecasting is also examined in this paper. And, the discrete wavelet decomposition is used to extract the trend components of vibration time series. Finally, the hybrid prediction model and algorithm of combining SVR and DWD is validated by a group of practical long-term vibration data measured from a flue gas turbine. The results show that the hybrid prediction model possesses more advantageous to forecast long-term state time series than directly using SVR model.