Fault Diagnosis Of Wind Turbine Drive Chain Based On Empirical Wavelet Transform

Wind turbines are generally serviced in harsh environments,affected by variable operating conditions and variable loads.Wind turbine generator chain systems,such as spindle bearings,gearbox gears,generator drives and free end bearings,etc.It is prone to failure,causing the unit to stop,which has a great impact on the service life of the wind turbine.In order to ensure the safe operation of the unit and improve the economics and market competitiveness of wind power generation,it is necessary to implement condition monitoring and fault diagnosis for key equipment of wind turbines.The Wind Turbine Condition Monitoring System(CMS)implements fault diagnosis by monitoring the vibration of the drive train components.After the vibration signal is obtained by the CMS system,it needs to be processed.However,due to the complicated operating conditions of the wind turbine,the measured vibration signal is seriously interfered by noise.In order to extract the weak fault signal under strong background noise,this paper studies the traditional empirical wavelet transform and the parameterless empirical wavelet transform based on scale space are applied to the vibration signal processing of wind turbine gearbox measuring point and bearing measuring point.Finally,the fault characteristic information of gear box and bearing is successfully extracted,and the validity of the method is verified.The main research contents and conclusions of this paper are as follows:(1)The paper studies the various failure modes of gearbox and bearing in wind turbine and the expression of its vibration signal in time domain and frequency domain,calculates the characteristic frequency of gear fault of fixed axle gearbox and planetary gearbox respectively,and calculates the four fault characteristic frequency of bearing.(2)The paper studies the traditional empirical wavelet transform,and the vibration data of gearbox test bench and the actual gearbox vibration data obtained from a wind turbine are processed,the defects of the components are successfully diagnosed,the fault location is positioned,and the effectiveness of the empirical wavelet transform in fault diagnosis is verified.Finally,compared with the effect of EEMD processing vibration signal,the superiority of empirical wavelet transfom in processing vibration signals is successfully verified.(3)The paper studies a parameterless empirical wavelet transform based on scale space,which has five methods.Through the processing of vibration data of gearbox test bench and actual vibration data of wind turbine gearbox measuring points,we can draw conclusions.The parameterless empirical wavelet transform based on scale space is more self-adaptive than the traditional empirical wavelet transform based on local maximum value,and the Empirical law method has advantages over other methods in the threshold Division based on scale space(4)The paper studies a feature extraction method based on Mahalanobis distance.which can quickly extract the fault-related AM components of all components obtained by the empirical wavelet transform based on the Empirical law.The method is used to process the vibration signals of the gearbox and bearing measuring points on the wind turbine drive chain,and finally successfully extract the imf containing the rich fault characteristics from the original vibration signal.And then successfully diagnosed the gear and bearing faults in the wind turbine,verified the effectiveness of the method.