Research On Feature Extraction Of Early Faults In Wind Turbine Gearbox

With the advancement of science and technology in our country as well as the development of the manufacturing industry,the demand for energy has been increasing.At present,the use of wind energy mainly relies on wind turbines,which can divided into doubly fed type,semi direct drive type and direct drive type according to driving system's structural characteristics.Among them,the double-fed type turbines are widely used due to advantages such as small size,relatively low cost and stable production of electricity.The key component of the double-fed model is the speedincreasing wind turbine gearbox whose operating conditions have a great impact on the entire turbine.Therefore,it is necessary to detect configuration and diagnose failure for speed-increasing wind turbine gearboxes.The purpose of the research is mainly to extract the fault characteristics of the gearbox vibration signal.Before the experiment,the rigid-flexible coupling dynamics simulation model of the secondary reduction gearbox was established based on the mapping dimensions.At the same time,the morphological filter & Hilbert-Huang transform method and the improved wavelet threshold denoising & Teager energy operator method were respectively used to process two groups of signals,which were separately generated by the WTDS experimental platform and its simulation model.During the experiment,the phenomenon of failure appeared when the fault features of the gear box were extracted by using the improved wavelet threshold denoising & Teager energy operator method.After analysis,the reason for the failure was that during the experiment,the fault and operation condition of the gear has been changed,which caused the fault characteristics of the gear to be covered,or offset by the other signals.To solve the problem,the signal was processed second times to obtain the characteristic frequency of the signal.It was found through experiments that the rigid flexible coupling model can basically simulate the actual situation.The fault feature frequency identification method,which contained the morphological filtering & Hilbert Huang transform method and the improved wavelet threshold denoising & Teager energy operator method,can extract the fault characteristic frequency of early tooth root crack and the early tooth surface wear occured on the WTDS platform secondary reduction gearbox input stage pinion.However,with the change of load and speed in the experiment,the failure rate of the Hilbert-Huang Transform extraction method based on morphological filter was improved.At the same time,due to the change of the gear fault status,different degrees of failure have occurred using the denoising & Teager energy operator extraction method based on the improved wavelet threshold.By improving the method based on improved wavelet threshold denoising & Teager energy operator,the success rate of early failure mode extraction of WTDS platform secondary reduction gearbox was improved.