Research On Fault Diagnosis Of Gears For Wind Turbine Gearboxes Under Random Wind Speed Conditions

The intelligent monitoring and health maintenance of wind turbines is an important guarantee for the development of wind energy.Wind turbine gearboxes are the core transmission components of wind turbines.which are prone to wear,cracks.broken teeth and other gear faults under long-term random wind speed conditions,and have great potential hazards.Therefore,the research on fault diagnosis of gearbox gears of wind turbines under random wind speed is in line with the need of safe and healthy development of wind power industry.However,due to the influence of random wind speed conditions,there are still some problems,such as unclear fault mechanism of wind turbine gearboxes and difficulty in extracting fault features of non-stationary signals with noise.In view of the above key problems,this paper takes the wind turbine gearbox as the research object,which is composed of planetary gear train and parallel shaft gear train.In order to achieve the research goal of gear fault diagnosis of wind turbine gearboxes under random wind speed,a vibration signal model of wind turbine gearboxes under random wind speed is established,and a non-stationary signal processing method based on PSO-EEMD and order analysis is proposed.Besides,the model simulation and experimental verification are carried out.(1)In order to reveal the mapping relationship between random wind speed,gear fault state and vibration signal response,a vibration signal model of wind turbine gearboxes under random wind speed is established.The random wind speed excitation is introduced into the vibration signal model for the first time,and the key factors such as time-varying meshing frequency of gears,time-varying fault characteristic frequency and planetary frame rotation effect are considered comprehensively.On the one hand,according to the statistical characteristics of wind speed,a random wind speed model based on Weibull distribution is established to simulate and generate random wind speed,which is then transformed into the input shaft speed of the gearbox.On the other hand,the time-varying fault characteristic frequency and carrier signal are obtained by combining the input shaft speed,structural parameters and gear fault status of the wind turbine gearbox.Then,the vibration signal model of the wind turbine gearbox is established through modulating the amplitude and frequency of the carrier signal by the fault characteristic frequency.Finally,through the simulation and analysis of the vibration signal model of the wind turbine gearbox under random wind speed,the non-stationary vibration signal containing the fault characteristic information of the gearbox affected by random wind speed is obtained,which provides theoretical basis and data support for the fault diagnosis of the wind turbine gearbox.(2)In order to improve the ensemble empirical mode decomposition(EEMD)method and play a better role in filtering and denoising,the PSO-EEMD method is put forward based on particle swarm optimization(PSO)to optimize the key parameters of the traditional EEMD method.This method chooses a new evaluation objective function of PSO to characterize the uniform distribution of the original signal extremum points,adaptively determines the optimal amplitude of the Gauss white noise Anoise,and calculates the corresponding total average number N,so as to improve the adaptability and decomposition accuracy of the EEMD method by optimizing these two key parameters.The simulation results show that the PSO-EEMD method can effectively decompose the original signal into multiple intrinsic mode functions(IMFs).Compared with the traditional EEMD method,the PSO-EEMD method is effective and superior in vibration signal filtering and denoising.(3)In order to extract fault features from non-stationary vibration signals of wind turbine gearboxes disturbed by noise,a non-stationary signal processing method based on PSO-EEMD and order analysis is proposed,which combines the advantages of PSO-EEMD in signal filtering and denoising.Firstly,the non-stationary time-domain signal of the wind turbine gearbox is transformed into the stationary angle-domain signal by means of equal angle sampling method.Secondly,the PSO-EEMD method is used to filter and denoise angle-domain signals of gearbox,and then the fault-sensitive IMF signals are screened according to kurtosis criterion.Finally,the order spectrum of the fault-sensitive IMF signal is obtained by Fourier transform,and the fault characteristic order of gears is extracted from the order spectrum to judge the fault state of wind turbine gearboxes.The simulation results show that this method can effectively extract the fault feature order of the non-stationary vibration signal disturbed by noise,and provide a technical support for the fault diagnosis of wind turbine gearboxes.(4)In order to verify the validity of the above models and methods,a case study of gear fault diagnosis of wind turbine gearboxes in a wind farm in China was carried out,aiming at the most common gear faults,such as gear broken fault and gear wear fault.Firstly,the scheme of collecting test data is described in detail.Besides the time-domain vibration signals of the wind turbine gearbox and the speed data of the output shaft of the gearbox are collected synchronously.Then,the non-stationary signal processing method based on PSO-EEMD and order analysis is used to obtain the order spectrum of fault-sensitive IMF signals.Finally,the fault feature order extracted from the order spectrum is used to directly and accurately judge the gear fault of the wind turbine gearbox,and it is verified with the actual test situation.