| In recent years,with the rapid development of China’s high-speed railways construction,the speed and mileage of trains have been continuously improved,which puts forward higher requirements for the safety of train operation.The gearbox bearing is one of the key component of the high-speed train transmission system.Once a fault occurs,it will directly threaten the safety of the train.Therefore,it is very important to diagnose the fault of gearbox bearing,especially to effectively detect and accurately diagnose the incipient stage of bearing fault.Because of the complex and harsh operating environment,the vibration signal component of the gearbox bearing is very complicated.When it appears incipient fault,the fault signal is weak,the signal-to-noise ratio is low,and the fault feature is difficult to be extracted.Therefore,it is the focus of this thesis to study how to effectively extract the incipient weak fault feature of gearbox bearings and realize the incipient fault diagnosis of gearbox bearings.In order to solve the problem that Teager Energy Operator is sensitive to noise,this thesis takes the strong anti-interference performance of the Frequency-Weighted Energy Operator(FWEO)demodulation and the advantage of the 1.5-dimensional spectrum depressing Gaussian noise and detecting the secondary phase coupling.A demodulation analysis method of 1.5-dimensional FWEO energy spectrum combining the two is adopted.Through the analysis and processing of the simulated signals,the properties of the two methods are illustrated,and the effectiveness and advantages of 1.5-dimensional FWEO energy spectrum is verified.The Maximum Correlated Kurtosis Deconvolution(MCKD)considers the periodicity of fault impulse component,which can enhance the continuous impulse component submerged in strong noise.Aiming at the problem that the deconvolution effect of the method depends on the parameter selection,the thesis introduces a global search optimization algorithm—Beetle Swarm Optimization algorithm,which is combined with the MCKD to optimize the search of the influence parameters(T,L and M),and realizes the adaptive MCKD.The combination of adaptive MCKD and 1.5-dimensional FWEO energy spectrum are applied to the simulated signal and bearing test data of high-speed train gearbox to verify the effectiveness of the method.Because MCKD can only extract part of the pulse components,and it is still iteratively solved with low computational efficiency,Mc Donald proposed a non-iterative deconvolution method,which is called Multipoint Optimal Minimum Entropy Deconvolution Adjusted(MOMEDA).The Singular Value Decomposition(SVD)is used as the prefilter to improve the multi-point kurtosis spectrum,which is difficult to effectively identify the weak fault periodic components in the strong noise environment.The variable step size search method is used to determine the optimal filter length,which avoids the uncertain factors of the artificial filter length.The combination of SVD-MOMEDA and 1.5-dimensional FWEO energy spectrum is applied to the simulation signal and the bearing test data of high-speed train gearbox,which verifie the effectiveness of the method,and shows that it can effectively enhance the impact component in the signal,realize the incipient fault diagnosis of bearing,and provide a new method for the incipient fault diagnosis of high-speed train gearbox bearing. |
