Vibration Signal Analysis And Fault Diagnosis Of Axial Flow Turbine Generator

During the operation of the hydro generator unit,the vibration is inevitable.The abnormal vibration of the unit causes the fatigue damage of the parts and components of the unit,and the operation state of the unit becomes worse,which is very adverse to the safe and stable operation of the unit.When the vibration amplitude is large,it will seriously affect the unit load distribution and grid security.According to the research,80% of the fault feature information of the hydro generator can be displayed in the vibration signal,but because of the complex operation environment of the hydro generator unit,the vibration data of the unit is affected by many factors during the collection process,resulting in the collected data affected by noise,abnormal pulse and other factors.It is very important for the safe,stable and economic operation of the power station and even the power grid to study how to remove the noise and abnormal pulse and other interference factors,accurately extract the vibration signal characteristics of the unit,judge the operation condition of the unit,and determine the appropriate maintenance time and work.First of all,the fault characteristics of hydroelectric generating units are analyzed in detail.According to the test data of variable speed,variable excitation and variable load test of the actual units of the hydropower station,combined with the trend curve characteristics,frequency spectrum characteristics,axis track characteristics and shafting state characteristics of the swing data,the causes of the unit vibration are found out.Through the analysis,it is found that the swing value of the rotating part of the unit is larger than that of other parts when doing these three tests;through the analysis of its trend curve characteristics,spectrum characteristics and axis track characteristics,it is found that its characteristics are the same as the mass imbalance of the rotating parts of the unit.Through the analysis of the shimmy data of other parts of the unit,it is shown that there are medium frequency pressure pulsation in the draft tube of the unit,uneven air gap or unbalanced magnetic potential in the rotor of the generator.Then,according to the non-stationary and non-linear characteristics of theshimmy data of the hydrogenerator unit and the problem that it is difficult to extract the fault features of the unit due to the noise and other characteristics,a fault diagnosis method of the vibration signal of the hydrogenerator unit based on the variable mode decomposition(VMD)and sample entropy is proposed.Firstly,VMD is used to decompose the data;then the sample entropy of each modal component is calculated to determine the selection threshold of sample entropy;finally,the selected modal component is reconstructed to achieve the purpose of denoising.The effectiveness of the method is verified by simulation analysis and examples,and the method is compared with LMD and ceemd methods.The results show that the signal-to-noise ratio of the denoised data is higher,the correlation coefficient is as high as 0.9937,and more effective information in the original signal is retained.The denoised data can accurately extract the fault characteristics of the group data.Then,according to the problem that the shimmy data of hydrogenerator unit contains noise and abnormal peak,and the signal is non-linear and non-stationary,which makes it difficult to extract the fault feature information,a fault feature extraction method of hydrogenerator unit based on non-linear mode decomposition(NMD)is proposed.The signal with noise and abnormal pulse is adaptively denoised and reconstructed by NMD,and the correlation between the reconstructed data and the original signal data is analyzed.The correlation coefficient is as high as 0.9908,which can effectively achieve the purpose of noise reduction and removal of abnormal spikes.The score analysis of NMD,EMD and ceemd shows that the data processed by NMD has fewer components,no redundant components and more At last,the validity and feasibility of NMD method in fault feature extraction of actual unit shimmy data are verified by an example.Finally,aiming at the problem that it is difficult to extract the characteristics of pressure fluctuation data of draft tube and other parts under the background of strong noise,an adaptive local iterative filtering(ALIF)method combining sample entropy and singular value decomposition(SVD)is proposed.Firstly,the data to be decomposed is decomposed by ALIF,then the components are selected to reconstruct according to the set sample entropy threshold,then the reconstructed signal is decomposed by singular value,and the reconstruction number is selected according to the position of the mutation point of singular value spectrum to reconstruct,so as to achieve the denoising effect.Comparing this method with EMD method,it is found that the denoising effect of this method is better.Through the simulation and example verification,it is found that this method can extract the feature signals under thebackground of strong noise accurately,which is very conducive to fault diagnosis.