Wind Turbine Gear Fault Diagnosis Based On Improved LMD And IRLS-SVM

Fly fierce development of wind power industry makes the wind turbines running stability and reliability of the gear need higher requirements.Wind turbines including gear box,generator,blades,hydraulic system and yaw system,gear box and generator is the main key components,but also the site of a greater incidence of failure,they are safe and stable operation will affect the performance of the entire unit,in order to reduce the incidence of wind turbine gearbox fault,improve operational eff-iciency,must seek to reduce the incidence of gearbox fault and reliable method.This article in view of the wind turbine gear vibration signal is non-stationary and time-varying characteristics,from unit fault mechanism and characteristics of the fault signal,the early fault diagnosis of wind turbine gear for this paper.Unit combines the gear box failure mechanism of the parts and units are summarized typical fault gear box form,with the improved LMD robust least squares support vector machine(SVM)and iterative method of combining the,by improving the local mean decomposition(LMD)improvement of wind turbine gear fault vibration signal is decomposed,then extract the characteristic parameters of energy,and its characteristic parameters were normalized processing,finally through iterative robust least squares support vector machine(IRLS-SVM)to different running condition of the gear unit(normal fault and broken teeth,wear failure)classification analysis..In this paper,the main content is summarized as:(1)The wind generator was introduced and discussed the mechanism and the typical fault of gearbox vibration,this paper introduces the principle of wind turbines and the gear box failure mode and fault vibration signal characteristics,structures,experimental platform,the related experiment data;(2)Introduced the basic principle and algorithm of LMD,main characteristics,the endpoint effect and energy feature extraction method,problems endpoint effect due to the local mean decomposition,using adaptive waveform matching improved method of continuation.Experiment and simulation results show that the improved local mean decomposition can effectively inhibit the endpoint effect;(3)Introduces the iterative robust least squares support vector machine(SVM)as well as the main characteristics,the basic principle and algorithm for fault identification problem of iterative robust least squares support vector machine(SVM)was applied to wind turbines gear fault diagnosis analysis,classifying the fault identification authentication;(4)Local wind turbine gear vibration signal test was improved after treatment mean decomposition,extraction of energy characteristic parameters,and for normalization of its energy characteristic parameters,and finally entered into the iterative robust least squares support vector machines classification.The method is simple,fault diagnosis of wind turbine gears is more effective.