Research On Dynamic Analysis And Fault Diagnosis Of Gearbox

Gearbox,as a transmission mechanism of torque,is widely used in the power transmission system of mechanical equipment.The gearbox is prone to malfunction because it has been in a bad working environment for a long time.Recognizing the working state of gearbox and estimating its damage degree can avoid accidents and reduce economic losses.Therefore,it is of great significance to recognize the working state of gearbox.In this paper,the gearbox is taken as the research object,and the simulation is carried out through dynamic analysis.According to the non-linear characteristics of the vibration signal of the gearbox,an improved empirical wavelet transform method is proposed,and the improved empirical wavelet transform and support vector machine are combined to identify the state of the gearbox.The main work of this paper is as follows:1.The engagement coincidence degree of two external gears in ideal state is calculated,and the meshing stiffness expression of the two external gears in the case of no fault is deduced,and the simulation analysis is carried out by using the single-degree-of-freedom dynamic model.The six-degree-of-freedom dynamic model is established,and the simulation results are analyzed in time and frequency domains.The results show that the vibration along the meshing line is more intense.Compared with the simulation results of single-degree-of-freedom dynamic model,the six-degree-of-freedom dynamic model is better than single-degree-of-freedom dynamic model.According to the expression of gear meshing stiffness without fault,the expression of gear meshing stiffness with a local spalling is deduced and simulated on the basis of six-degree-of-freedom dynamic model.The simulation results are analyzed in time domain and frequency domain for different degree of damage.The changing trend of kurtosis,root mean square,impulse factor and peak factor with the degree of gear failure is analyzed.It can provide reference for the follow-up work.2.The original empirical wavelet transform is used to segment the spectrum,and the spectral kurtosis,impulse factor and peak factor of each frequency band are extracted.The three indexes are fused by the entropy weight method to construct the revenue function,which is applied to the fusion of adjacent frequency bands according to the fusion criterion.The original signal is decomposed into a series of singlecomponents by the filter banks obtained from the spectrum fusion.The single component with the highest income growth rate is selected as Hilbert transform,and the envelope signal is analyzed by frequency spectrum.The improved method is applied to the simulation signal and experimental signal of the local fault of gearbox gears,which proves that the improved algorithm can detect the impulse component adaptively in the noisy signal.Compared with the original method,the improved algorithm is proved to be superior to the original algorithm,and can be used to construct the sample points of support vector machine in Chapter 4.3.The method of time-domain synchronous average de-noising is used to pre-process the collected vibration signals,which improves the signal-to-noise ratio.Empirical wavelet transform is used to divide the signal spectrum,extract the energy entropy of each frequency band to construct the signal,and combine the feature attributes extracted by time-domain statistical method to construct the feature vector of the signal as the sample points of support vector machine.The improved Support Vector Machine is trained and tested with three kinds of sample points,which can effectively identify the gear state.