Research On Fault Diagnosis And Condition Monitoring Of Wind Turbine Gearbox Based On Improved Ensemble Learning Algorithm

With the rapid development of the global new energy technology and the rising of the proportion of wind power,the importance and urgency of the research on the fault diagnosis and condition monitoring method of the wind turbine is more and more significant.The wind turbine gearbox fault diagnosis and condition monitoring method is the focus of the research.In order to improve the reliability of the wind turbine gearbox,gears,bearings and oil temperature are studied respectively.Centering on the combination of ensemble learning and artificial bee colony algorithm and an improved artificial bee colony algorithm,to study the wind turbine gearbox fault diagnosis and condition monitoring.Then software and hardware of wind turbine fault diagnosis and condition monitoring system are designed based on researched methods.Four aspects of work are done as follow.(1)For researching wind turbine gearbox fault formation mechanism,collection of gear and bearing vibration signal on the experimental platform is used.First of all,vibration signal is denoised through the wavelet package transform method and time domain features are extracted from it.Then signal in time domain is transformed into frequency domain signal through Fast Fourier Transform and frequency domain features are extracted from it.Finally,the value of time domain and frequency domain features are normalized and will be used in following chapter for fault diagnosis.(2)For diagnosing wind turbine gearbox gear pitting,gear broken,bearing inner ring damage and bearing outer ring damage state,selective neural network integrated algorithm based on artificial bee colony algorithm(ABCSEN)is proposed for fault diagnosis.Firstly,the UCI data sets show that ABCSEN is better than GASEN and Bagging in diagnosis accuracy and efficiency.Then history fault data of gearbox is used to train ABCSEN to obtain fault diagnosis model.Finally,by using new fault data to test fault diagnosis model,the results show that the model has good effects on diagnosis.(3)For researching gearbox condition monitoring method,the improved ABCSEN,namely selective neural network ensemble algorithm based on modified Cauchy bee colony algorithm(MCABCSEN)is proposed firstly based on wind turbine gearbox oil temperature data.Then test function is used to verify the performance superiority of improved algorithms.Finally,no.7 wind turbine gearbox oil temperature data in south wind field and the artificial fitting gearbox fault oil temperature data is used to train and test the new algorithm.It shows that the new algorithm is sensitive and can warn failures in advance,remind the staff timely and prevent further loss.(4)Based on the research of fault diagnosis and condition monitoring method,the wind turbine gearbox fault diagnosis and condition monitoring system is built,the design of system hardware and software is explained in detail and the design scheme is simply experimental verified finally.