Fault Feature Extraction And Diagnosis Of Wind Turbine Gearbox Under Variable Condition

The gear box is the main driving part of the wind turbine.The alternating load caused by the poor working environment and the change of the wind condition is very easy to cause the gear fault.The vibration signal shows strong nonlinear and non-stationary.It has important scientific significance and research value for the stable operation and efficient maintenance of the wind turbine that makes the best of the instantaneous vibration information under the variable working condition for the state monitoring and fault diagnosis.On the basis of the problems in the gear box fault diagnosis under variable working condition,this paper puts forward the feature extraction and diagnosis method under the variable working condition based on the image processing method.The result is carried out in the wind turbine gear box test table shows that the method can effectively extract the fault features of the gear box under the variable condition.The fault type recognition result is good.Firstly,this paper studies the method of generating the fault image under the variable condition.Through the comparison of the principle analysis and the simulation experiment,it points out the superiority of the time frequency analysis method to generate the fault image under the variable condition.Secondly,this paper studies different image feature extraction methods and points out the good performance of TD-2DPCA in image feature extraction.On this basis,the combination of three time-frequency analysis methods about STFT,WT,S transform and TD-2DPCA algorithm is analyzed,and the features from three time-frequency images is extracted by TD-2DPCA,and the support vector machine is used for fault diagnosis.It is found that the combination of wavelet transform with Morlet wavelet and TD-2DPCA is superior to S transform,and the combination performance of S transform is better than that of STFT,whether it is from the different dimensins feature extraction or the random experiment.Finally,the combination of WT and TD-2DPCA is used to extract fault characteristics under variable conditions in gear box.In different dimension feature extraction experiments,a higher recognition accuracy is achieved,up to 96%.Under the variable conditions,the recognition accuracy of TD-2DPCA is better than the 2DPCA.Compared with the PCA,the performance of the two dimensional PCA for extracting the fault image features of the gear box with variable working conditions is superior and the fault diagnosis accuracy is higher.In order to analyze the comprehensiveness of TD-2DPCA in extracting image features,it is compared with GLCM,LBP and Gabor image feature extraction algorithms.Through random experiments,it is found that TD-2DPCA has better characteristics in extracting image features.