Fault Diagnosis Of Gearbox Bearing Of Wind Turbine Based On EWT And Optimal Parameters Refined Compound Multiscale Dispersion Entropy

Wind is a renewable clean energy.The development of contemporary society consumes a lot of energy and energy is insufricient.The use of wind energy can effectively alleviate the problem of traditional energy shortage and solve the problem of environmental pollution caused by traditional energy.With the increasing cumulative installed capacity of wind turbines and the bad operating environment of wind turbines,wind turbine Gearbox failures occur frequently.Once the gearbox fails,the unit will face long downtime and expensive maintenance costs,resulting in huge economic losses.Therefore,accurate and efficient condition monitoring and fault diagnosis of the gearbox is of great significance to ensure the safe and stable operation of the unit and improve the power generation efficiency.This paper proposes a method based on EWT(empirical wavelet transform)and optimal parameter refine composite multi-scale dispersion entropy to study the fault diagnosis of wind turbine gearboxFirst of all,under actual working conditions,due to severe environmental noise interference,complex vibration signal transmission paths and electromechanical coupling,the vibration signal of the wind turbine gearbox has the characteristics of non-stationary、non-linear,and low signal-to-noise ratio.It is difficult to directly study the original vibration and extract effective fault information from the signal.In this paper,empirical wavelet transform is introduced to process the vibration signal of the wind turbine gearbox,and the submodal components are filtered through the correlation coefficient threshold to reconstruct the signal,so as to obtain a fault vibration signal with a higher signal-to-noise ratio.Compared with the EMD(empirical mode decomposition)method,Experimental results show that EWT can effectively extract the main components of the signal in the noise environment,which lays the foundation for the subsequent feature extraction.Secondly,for the feature extraction link,The traditional time-domain and frequency-domain fault feature extraction effect is not good,and the feature matrix has the characteristics of redundancy,resulting in the problem of poor fault diagnosis effect.In order to improve the performance of fault feature extraction,a new time-frequency feature refined composite multi-scale distribution entropy is introduced as feature vector,the square function of its skewness value is used as the fitness function and synchronized by the grid search algorithm Search and calculate the two key parameters m and C,extract the optimal parameters of the gearbox fault vibration signal,and construct the eigenvector matrix with refined composite multiscale entropy.Through experimental comparison,it is proved that the refined composite multiscale dispersive entropy with the optimal parameters of the EWT reconstructed signal can distinguish better when extracting various fault features,and the diagnosis result is more stable and accurate.Finally,the problem of feature vector redundancy and general classification algorithm has many parameters and the parameter setting affects the classification accuracy.The Relief-F algorithm is used to calculate the classification weight of the feature vector,and the one with the highest weight is selected to form the final feature vector,and redundant features are eliminated.Finally,the extreme learning machine with fast calculation speed and few parameter settings used for fault diagnosis.Through experimental analysis and comparison with other methods,it is proved that this method is more accurate and more stable,and can be effectively applied to the fault diagnosis of wind turbine gearboxes.It has certain value in practical engineering applications and is useful for the diagnosis of wind turbine gearbox faults.The related researches have a certain reference.