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

In this paper,through the understanding and understanding of the research results of signal feature extraction and fault classification and identification in domestic and foreign research scholars in recent years,combined with the research field of their own,it is determined that the working environment of large wind turbine generator sets is prone to failure,and the fault repair is difficult,and a problem that causes huge economic losses.The feature extraction and fault identification of the fan operating state signal are proposed to solve the shortcomings of traditional fan fault diagnosis with high labor cost,long periodicity and poor real-time performance.The structure and components of the wind turbine are analyzed in detail,and the gear transmission part is analyzed.The common fault types of the gear transmission system are classified and summarized,and the possible fault causes of various faults are discussed..The gear ratio of the gear transmission system and the rotational frequency of each gear are calculated,and the fault characteristic frequency of each gear can be determined,which plays a decisive role in guiding the frequency domain analysis of the fault situation later.A brief introduction to the time domain analysis and the frequency domain analysis method.The time domain analysis method is to calculate some parameters of the signal for a certain length of time,and these statistical values are indicators;the frequency domain analysis is to transform the time domain signal.To the frequency domain expression,the characteristics of the frequency domain that the signal is changed are analyzed.Then through the experimental data is the signal analysis,the statistical analysis of its time domain statistical indicators,can explain the characteristics of the signal in the time domain and frequency domain,can be used as a reference indicator for fault diagnosis,but also exist obvious defects provide an indispensable theoretical basis for the subsequent chapters of research.The theoretical analysis of the two types of mainstream recursive modal decomposition methods(EMD,LMD)and the algorithm flow are analyzed in detail.The theoretical analysis of the end effect and modal aliasing disadvantages of the two types of methods is carried out.The inevitable inherent in the two methodologies.The introduced variational modal method(VMD)solves the modal aliasing effect well.Therefore,the VMD signal decompositionmethod is used to decompose the experimental signal,and the eigenvalues of the components are determined in the form of energy ratio,so that the extracted features are more vivid,providing powerful data for fault diagnosis of feature recognition by feature data.in accordance with.Various popular classification and recognition models are analyzed,and then the fuzzy neural network is selected as the model method for fault classification and identification based on the characteristics of classification data.The VMD signal decomposition method is used to decompose the normal state and the fault state,and then the energy ratio of the decomposed signal is used as the fault feature of fault identification.The fault feature with more obvious classification is used as the fuzzy neural network.Input parameters,through the evaluation of the classification accuracy rate,can be sure that the fuzzy neural network achieves the ideal classification effect in fault classification.