Research On Vibration Signal Feature Extraction And Fault Diagnosis Of Planetary Gearbox

The planetary gearbox is widely used in the complex transmission system such as the helicopter,the wind turbine,and so on.It has small volume,high power to weight ratio,bad environment and complex working conditions,which leads to the large probability of fault occurrence.It is of great significance and economic benefits to realize the fault diagnosis and early warning of the planetary gear box.In this paper,the fault feature extraction and fault diagnosis of planetary gearbox are studied,the main research contents are as follows:(1)Analysis of vibration characteristics of typical planetary gear damage condition.Aiming at the common failure mode of planetary gear,and analyzed the formation mechanism and the reasons,establish two stage planetary gear dynamics model of rigid flexible multibody system.The characteristics of the vibration signals of the second planetary gears are analyzed,such as variable speed,variable load,broken teeth and cracks.(2)The fault diagnosis method based on empirical wavelet transform based on energy aggregation is studied.According to the planetary gearbox early fault,fault characteristics of weak signals of defects in the experience of wavelet transform method and the lack of research on new energy aggregation experience based on wavelet transform and adaptive signal decomposition method,the simulation and experimental results show that the new method can maximize the retention of the vibration signal sidebands,the early fault of the sun gear is detected effectively.(3)A method for the separation and fault diagnosis of planetary gear coupling vibration signals based on total variation model and basis pursuit algorithm is studied.Aiming at the complex planetary gear vibration signals,the complex planetary gear vibration signal is divided into impulse modulation and steady-state modulation components,and constructs the model,using the sparse total variation model in rotating mechanical shock signal feature extraction,using the improvement asymmetric punishment function,Majorization-Minimization method are used to solve the model.Separation of steady state modulation components by using the basis pursuit method.By analyzing the extracted characteristic signal,the fault information is identified.Simulation and experiment prove the feasibility and effectiveness of the new method.