Application Of Improved DTW Algorithm In Fault Diagnosis Of Wind Turbine Transmission Gear

In recent years,with the aggravation of energy and environmental problems,green power generation technology represented by wind power generation has attracted more and more attention,and the installed capacity of wind turbines continues to grow.However,because wind turbines usually operate in complex and changeable harsh environment,wind turbines are prone to induce various types of failures,especially for early faults,which are difficult to be found and dealt with in time.With the continuous operation of wind turbines,early faults will gradually aggravate,and even lead to major safety accidents in serious cases.Gear is a faultprone component in wind turbines.When it fails,it is difficult to repair and replace,and it will lead to the wind turbine being shut down for a long time.Therefore,strengthening the research on early fault diagnosis of wind turbine gears and achieving early fault detection and early treatment is of great significance to prolong the service life of wind turbines and improve power generation efficiency.Taking the transmission gear in the wind turbine gearbox as the research object,the early local fault diagnosis of the transmission gear is studied using the vibration analysis method in this thesis.Because the early local fault characteristics of gears are relatively weak,they are easily affected by background noise,and similar to the characteristics of normal gears,which is easy to be misdiagnosed.In addition,with the development of wind turbines towards largescale and complexity,the components of gear vibration signals collected by vibration sensors have become increasingly complex,and it is difficult to effectively identify early faults of gears only based on traditional time-domain analysis and frequency-domain analysis of vibration signals.Therefore,the main research works of this thesis are as follows.(1)Analysis of gear vibration characteristics from failure mechanism.Firstly,the simplified meshing model of single-stage spur gear and the discrete dynamics model of gear meshing are established.The components of vibration signal generated in the meshing process of gear pair and the dynamic characteristics of gear in each period of power transmission are analyzed.At the same time,several common types of gear faults are introduced,and the manifestations of gears in different states on the frequency spectrum are given.Finally,the difficult problems of fan gear vibration fault diagnosis are given.(2)The traditional DTW algorithm will form singularities in the process of matching sequences.To address this problem,an improved DTW(IDTW)algorithm is proposed.The IDTW combines the angular similarity and Euclidean distance to construct a new eigenvalue,which is used to match the corresponding components of the two sequences.Through comparative analysis,the result shows that the IDTW algorithm can effectively solve the singularity problem of traditional DTW algorithm,and has high analysis accuracy.(3)Combining the IDTW algorithm with multivariate variational modal decomposition(MVMD),cross-correlation and resampling techniques,a method for extracting early fault features of gears based on MVMD and IDTW is proposed.It is used to remove the normal state components contained in the early local fault signals of gears,so as to effectively highlight the early fault features of gear.The analysis results of the simulation and test data of the early local faults of the gears show that the method can effectively extract the early local fault characteristics of the gears.(4)Aiming at the problem of fault pattern recognition of actual gears,a combination of IDTW algorithm and K-nearest neighbor classification algorithm is proposed and applied to the fault recognition of transmission gears in wind turbine gearbox.In this method,the shortest distance value obtained in the IDTW algorithm as the basis for judging the similarity between the test data set and the training data set,and realizes the classification of the gear test data set according to the principle of the K-nearest neighbor classification algorithm.Through the analysis of the actual gear data of the wind turbine drivetrain diagnostic simulator,the results demonstrate that this method can effectively realize the identification of gear faults in different states,and thus has high classification accuracy.