Research And Application Of Generator Fault Diagnosis Based On Large-Scale Wind Turbine

By the end of 2017,China's total installed capacity of wind power has reached a new high,reaching 195167 MW,ranking first in the world.But with its rapid development,the failure of wind turbines is also increasing,which directly affects the normal operation of the whole machine and even the whole wind farm.As an important device to transform wind energy into electric energy,generator not only affects the efficiency of wind power conversion system,but also affects the quality of output power.The bearing is the key part of the generator's rotating part in the fan,which is very important for its normal operation.Therefore,the research of generator fault diagnosis is of great value for improving the safety and reliability of wind farm operation and reducing maintenance rate.This paper starts with the structure of doubly-fed wind turbines and permanent magnet wind turbines,compares and analyzes the different working principles of the two wind turbine generators,and uses the two types of large-scale intelligent operation and maintenance systems independently researched and developed by this research team.The three-phase stator current of the generator is studied on the operating state of its generator.The Hilbert-Huang Transform(HHT)is used to analyze the doubly-fed asynchronous generator.The fault frequency of the stator and rotor of the generator can be calculated by the motor current method.With frequency,it can be determined whether the generator is in good operating condition.The analysis of field data confirmed the effectiveness of the method.According to the operating characteristics of permanent magnet wind turbines,it is known that there is no slip rate.This paper proposes Park vector transformation for the analysis of permanent magnet synchronous generators.According to its principle,it can be seen that the three-phase stator current of the generator is brought into Park.In the vector transformation,you get a graph and a polar graph.If both figures are near-circular,it indicates that the generator is operating normally.If both graphs are irregularly closed curves,it indicates that the operation state is not good,and it can be determined according to the angle corresponding to the long axis in the polar coordinates.This method was applied to the scene and confirmed the correctness of the method.For the analysis and diagnosis of the bearing state in the generator,a method of combining local mean decomposition(LMD)with improved spectral kurtosis was proposed.The improvement of the spectral kurtosis based on the wavelet packet decomposition method overcomes the defect that the window function variation in the original spectral kurtosis is limited.The original signal is decomposed by local mean value,and the PF component with a large correlation coefficient is selected for reconstruction.By improving the spectral kurtosis analysis,the band with the largest kurtosis of the reconstructed signal is taken as the bandpass filter,and the envelope demodulation is performed after filtering.The envelope signal containing bearing fault information can be obtained,and the demodulation of the fault signal can be achieved perfectly,eliminating the modal aliasing phenomenon of local mean decomposition.Through the analysis of the simulation signal and the verification of the test bench,the accuracy of the method was verified.Finally,this method was applied to practical engineering of wind farms.The fault diagnosis of a wind turbine was realized and the fault was accurately located,which provided a new idea for fault diagnosis research of transmission system.This article also improved the frequency spectrum analysis of generators and generator bearings in the system,completed the time-frequency analysis of vibration signals and current signals,and performed diagnostic functions for generators and bearings on the basis of the original intelligent operation and maintenance system.Addition,the application of the generator fault diagnosis method is realized,which provides a convenient way for the state analysis of the generator in the wind power field.