Research On Feature Extraction And Fault Diagnosis Method Of Wind Turbine Gearbox

In today's society where energy shortages and environmental pollution are becoming more and more serious,countries around the world are increasingly attaching importance to clean energy.As a clean energy source,wind energy has also received widespread attention.However,the working conditions of wind turbines are severe and prone to sudden changes,which makes the unit equipment extremely prone to various failures.For this purpose,this paper introduced Complete Ensemble Empirical Mode Decomposition with Adaptive Noise(CEEMDAN)into the gearbox fault diagnosis of wind turbines,and studied the fault feature extraction of wind turbines based on CEEMDAN.A fault diagnosis model of wind turbine generator based on Improved Particle Swarm Optimization Support Vector Machine(SVM)algorithm was proposed to realize the accurate identification and diagnosis of wind turbine faults.The main research contents and research results of this paper are as follows-introduced the basic structure and power generation principle of wind turbines,analyzed the common fault types and fault causes of wind turbine gearboxes,and introduced the processing methods of gearbox signals for subsequent wind turbine detection and faults.This paper introduced the theory of time-frequency analysis method,introduced the empirical mode decomposition method,focused on the Hilbert-Huang Transform(HHT)analysis method,and introduced the empirical mode.Theoretical analysis was performed by Ensemble Empirical Mode Decomposition(EEMD)and CEEMDAN algorithm.Through the simulation of Matlab analog signal,the superiority of HHT time-frequency analysis method in analyzing non-stationary state signals was expounded in detail,and the indicators ofmeasured signals were objectively evaluated.The simulation results showed that the improved CEEMDAN algorithm can effectively suppress the modal aliasing phenomenon and has a good decomposition effect.The experimental device and data acquisition device used were introduced.By replacing the faulty gear,the fault state of the single fault or composite fault of the simulated gearbox was realized.Through the analysis and processing of the collected vibration signals,the fast and accurate processing method was selected,and the time-frequency domain information of the gearbox failure was extracted to realize simple fault diagnosis.The energy entropy method was used to process the eigenmode function component obtained by CEEMDAN decomposition,and obtained the feature recognition vector of the vibration signal.A fault diagnosis model based on improved particle swarm optimization algorithm for support vector machine was established.The fault diagnosis method based on CEEMDAN-IPSO-SVM was applied to the fault diagnosis of wind turbines.The vibration fault data of the gearbox experimental platform was selected to identify and diagnose the vibration faults of the unit,and compared with EEMD-IPSO-SVM.It was proved in this experiment that the method has high diagnostic accuracy and realizes the effective diagnosis of wind turbine gearbox failure.