Research On Fault Diagnosis Of Wind Turbine Bearings Based On Symplectic Geometric Mode Decomposition And Adaboost Algorithm

In order to implement the new development concept and build a clean and beautiful world,together with the "carbon peak" and "carbon neutral" goals,the development and utilization of clean energy has been greatly accelerated.Wind power,as one of the clean energy sources,has been increasing year by year in installed capacity in recent years,and the domestic market has continued to expand.Its competitive advantages in the energy industry have become more apparent.Because the service environment of wind turbines is harsher and the structure is more complicated,this puts forward higher requirements for its operation and maintenance.Rolling bearings are an important component of wind turbines,and their health status has an important impact on the normal operation of wind turbine.Therefore,it is of great significance to carry out fault diagnosis for the rolling bearings of wind turbines.Based on this,this paper takes rolling bearings as the research object,and proposes a wind turbine bearing fault diagnosis method based on symplectic geometric mode decomposition and AdaBoost algorithm.The main work of this thesis is as follows:Firstly,study the processing method of nonlinear vibration signals.Symplectic Geometric Mode Decomposition(SGMD)is a newly proposed signal processing method.Because of its superiority,it has attracted more and more attention in the field of fault diagnosis.However,the similar component recombination problem involved in this method has not been stated clearly.To solve this problem,this paper introduces cosine similarity(CS)on the basis of the SGMD algorithm,and improves the SGMD-CS method.On the one hand,it can distinguish the effective component and the noise component,and realize the noise reduction of the original signal.On the other hand,the initial components with similar characteristics are reconstructed to obtain the final trend components.By constructing the simulation signal without noise and with noise,and comparing it with the decomposition performance of other signal processing algorithms such as EMD,LMD and VMD,it shows that SGMD-CS has good decomposition performance in processing complex signals,and has a strong robustness,so this algorithm is used to decompose and process the vibration signals of rolling bearings.Secondly,the feature vector extraction is studied.Based on the SGMD-CS algorithm,the symplectic geometric entropy(SymEn)is calculated by combining the components obtained from SGMD with the information entropy,which is used as the fault feature vector.For the purpose of verifying the superiority of SymEn,the fault classification results obtained by using approximate entropy,sample entropy and fuzzy entropy as feature vectors were compared respectively,and the accuracy,precision,recall and F1 score were used as evaluation indicators for each model.The classification results show that,compared with the other three kinds of entropy,the SymEn can effectively extract the fault information and make the diagnosis result of the classifier better.Finally,the fault diagnosis model of wind turbines is studied.Based on the aforementioned signal preprocessing and signal feature extraction,in order to achieve pattern recognition,the idea of ensemble learning is adopted to assemble multiple weak classifiers into a strong classifier,so as to improve the accuracy of fault classification.This paper chooses the AdaBoost algorithm,gives a bearing fault diagnosis model based on SGMD-CS and AdaBoost algorithm,and verifies the validity of the model through the vibration data of the actual rolling bearings.