| At the beginning of the 14 th Five-Year Plan,"high quality" and "high efficiency" have become new development themes,and new requirements have been put forward for industrial development.However,once mechanical equipment fails,it will inevitably affect industrial production efficiency and product quality,and even cause catastrophic accidents,which will have a negative impact on society and enterprises.Gearboxes are responsible for transmitting motion and torque,which are the important components in mechanical equipment and widely used in wind power,chemical industry,transportation,aerospace and other fields.Gearbox is one of the parts that are prone to failure in mechanical equipment due to the harsh working environment and the variable load.In this dissertation,we take gears and bearings as research objects and deeply study their failure mechanisms and sparse decomposition diagnosis methods,which will provide basis and methods for the source prevention of gearbox faults and effective fault diagnosis.We study the fault diagnosis technology of gearbox from two aspects: dynamic simulation and fault data processing.First,based on Workbench and Lsdyna,the steady-state and transient finite element simulation models of gears and bearings are established.The fault mechanism,fault dynamic response and fault signal characteristics are studied through model analysis.Then,the gearbox fault vibration data acquisition experiments are carried out.Based on experimental data,we study how to apply sparse decomposition to fault diagnosis of gears and bearings and feature extraction.Simulation and experiment are based on each other and are compared with each other.The content of the paper mainly includes the following four parts:(1)Based on Workbench,the modal analysis and harmonic response analysis models of gears and bearings under normal conditions and different fault conditions are established respectively.According to the simulation models,the effects caused by different defects on the natural frequency,modal formation,resonance phenomenon and main vibration direction of the part are analyzed.Furthermore,the influence of the fault on the main vibration direction of the part’s mode shape is analyzed.(2)Based on Lsdyna,the explicit dynamic models of gears and bearings under normal conditions and different fault conditions are established respectively.According to the dynamic simulation results,the influence of the fault on the dynamic response of the gear pair and the bearing system is studied;the time domain and frequency domain characteristics of the vibration acceleration signal of the faulty part and the normal part are analyzed.At the same time,the stress concentration phenomenon caused by the failure is analyzed.The results obtained by the simulation will provide certain guidance for the study of the fault mechanism and fault diagnosis technology of gears and bearings.(3)An over-complete fixed dictionary construction method combining dual-tree complex wavelet decomposition and Laplacian correlation filtering is proposed to improve the correlation between dictionary atoms and signals.This method improved the correlation between the dictionary atoms and the signal and reduced the parameter range of the generated complete dictionary which made the sparse decomposition better applied to the fault diagnosis of gears.First,the anti-aliasing and translation invariance characteristics of dual-tree complex wavelet decomposition are used to separate different signal sources in the original signal.Then,the Laplace wavelet feature library is used to perform correlation filtering on the decomposed signal and identify the characteristic parameters of periodic impulsive feature.According to the parameters,a complete dictionary is constructed,and the sparse reconstruction of the fault features in the signal is realized based on the dictionary.Finally,the method is applied to the vibration signal processing of single-shaft gear faults and double-shaft gear faults,and the proposed method effectively realizes the diagnosis of gear faults.(4)An improved sparse decomposition algorithm based on genetic algorithm and envelope entropy is proposed and applied to bearing fault diagnosis.First,it is proposed to combine dual-tree complex wavelet decomposition and kurtosis to extract the optimal component containing the impact feature to achieve the initial noise reduction of the signal.Then,the matched tracking improved by genetic algorithm is used to reconstruct the processed signal,and the impact feature of the signal is further extracted to realize the deep noise reduction of the signal.Aiming at the problem that it is difficult to set the termination condition in sparse decomposition,a termination criterion based on the the residual signal envelope entropy is proposed.Finally,the method is applied to the diagnosis of different bearing faults.The experimental results demonstrate that,compared with the traditional sparse decomposition,the proposed method has stronger feature extraction ability and can accurately identify different faults of the bearing when dealing with strong noise. |
