VMD-HT-ResNet-BASED VIBRATION CONDITION MONITORING AND EARLY WARNING FOR WIND TURBINE TOWER
Aiming at the problem that it is difficult to reliably monitor and warn the multimodal vibration of large wind turbine tower,a multi-input and multi-output(MIMO)nonlinear dynamic modelling method combining variational modal decomposition(VMD)-Hilbert transform(HT)-residual network(RestNet)is proposed to model the nonlinear dynamic characteristics of tower vibration of wind turbines.The inherent modal vibration of the tower is decomposed and extracted,and the differential dynamic regression vectors that accurately characterize the vibration characteristics of the tower are defined,and balanced sampling and MIMO modelling are carried out for the whole working conditions.Then,an exponential moving average(EWMA)index is established to evaluate the health of the tower,which is used for its condition monitoring and early warning.Simulation results show that the proposed method achieves high-precision monitoring and early warning of the tower vibration characteristics of wind turbines under all operating conditions,which provides a guarantee for the high safety operation of the turbines.
NETL
NSTL
万方数据
