Research On The Fault Characteristic Analysis And Diagnosis Method Of Wind Turbine Gearbox

With the development of society,the global energy shortage,environmental pollution and other social problems have become increasingly prominent.Because of its rich reserves,clean and environmental protection,renewable and other advantages,wind energy has become one of the key energy developments in each country.According to the study,the wind turbine transmission system fault occurred frequently,and the loss is the most serious,so it is particularly important to study the gear box which is the key parts of the wind turbine transmission system.Aiming at the gear box gears and bearing failures,we mainly do the following work in this paper.(1)The paper studies the gearbox fault mechanism of large-scale wind turbine and introduces fault diagnosis method based on vibration signal.The gearbox vibration signal acquisition often contains a lot of noisy information,resulting in a great impact on fault recognition.This paper uses the method of combination of morphological filter and noise reduction method that can effectively reduce the noise.The method compared with common wavelet threshold de-noising method and wavelet packet de-noising method.Meanwhile,the fault diagnosis cases of gear box are analyzed.The results show that by using combined morphological filtering and noise reduction method,the vibration signal has a higher signal to noise ratio and a smaller mean square error,and the method has better effect.(2)Fault feature extraction and fault diagnosis of gear box vibration signal of large wind turbine generator are studied.In this paper,energy entropy is introduced as the characteristic parameter of fault diagnosis,and the new method of vibration signal feature extraction based on ensemble empirical mode decomposition(EEMD)-energy entropy is proposed.The EEMD can availably solve the problem of mode mixing caused by traditional empirical mode decomposition(EMD)by adding the Gauss white noise.Compared with wavelet packet energy entropy feature extraction method,it has adaptive capacity and higher resolution ratios in both time domain and frequency domain.(3)Aiming at the nonlinear and non-stationary signal characteristics of the wind turbine gearbox vibration signal,this paper presents a BP neural network fault diagnosis method based on ensemble empirical mode decomposition and improved quantum genetic algorithm.EEMD is used to decompose the vibration signal sequence into a series of intrinsic mode functions(IMFs).Feature parameters are extracted.The dimension of the feature parameters set is reduced with principal component analysis.A training neural network for the feature parameters set is used to establish a fault diagnosis model.In the training process,improved quantum genetic algorithm is used to optimize weights and thresholds of the BP neural network.Experimental results show that the above mentioned model can effectively extract main features of the vibration signal and identify faults of the gearbox.