Research On Fault Diagnosis Method Of Wind Turbine Gearbox

As a clean renewable energy source,wind energy is used world wide nowadays.The installed capacity of wind turbines has increased dramatically,and the failure has occurred frequently,which will cause irreparable damage.Therefore,its fault diagnosis is imminent.The gearbox of the wind turbine has a high frequency of failure and its maintenance is difficult.Therefore,this paper analyzes and studies the fault diagnosis method of the fan gearbox.The main work has the following points:(1)Analyzed the basic principle and common fault mechanism of the fan gearbox,and summarized the research status of its fault diagnosis.(2)An eigenvalue extraction method combining improved wavelet threshold denoising and improved empirical mode decomposition(EMD)is proposed to extract the noise-free and more accurate time domain features and each eigenmode obtained by decomposition.The fault characteristics of the state function(IMF)component energy form a sample set.(3)Aiming at the redundancy of eigenvalues in the sample set,an improved neighborhood rough set method is proposed to reduce the eigenvalues that have no influence on the decision attributes.The superiority of the method is verified by experimental analysis of the UCI data set.(4)For the fault of the fan gearbox,an improved particle swarm optimization algorithm(IPSO)was put forward to improve the BP neural network,and then establish the IPSO-BP neural network model.Through the comparative analysis of the experiment,the superiority of the proposedmethod in the fault diagnosis of the gearbox is proved.(5)Set up the experimental platform and design a gearbox fault diagnosis software system with Matlab GUI interface.Integrated signal acquisition processing,feature value extraction reduction,fault diagnosis and other functions.Make the diagnosis work more intuitive and simple.