Gear transmission system is widely used in energy equipment,aerospace,industrial robots and other fields because of its advantages of high transmission accuracy and high operation stability.However,due to the influence of random working conditions,multi-source external mechanism and structural complexity,many key components are prone to different degrees of damage in the whole life cycle.In order to ensure the safety and working stability of the equipment,the research on the life degradation and fault diagnosis of the key components in the gear transmission system has been a hot topic for scholars at home and abroad.In the past research,scholars have relatively little research on the shaft crack fault of gear transmission system.However,as the key component of transmission system to transfer torque and load,transmission shaft is prone to failure.In order to avoid the incalculable loss caused by small loss,it is necessary to carry out deterioration mechanism and fault diagnosis.At the same time,in the modeling of transmission system,the flexible effect of transmission shaft is generally ignored.According to the research results,the model accuracy can be further improved by properly taking into account the flexibility of the transmission shaft,and the modal changes can be effectively characterized.It has certain engineering value and significance to carry out the flexible modeling and fault diagnosis of gear transmission system with shaft cracks.In view of the above,this paper mainly carried out the following work:First of all,this paper takes the transmission system of the two-stage cylindrical spur gear reducer of wind turbine as the research object,combines with the finite element idea,effectively takes into account the flexibility of the transmission shaft,and establishes the finite element dynamic model of the transmission system.On this basis,the stiffness reduction matrix of shaft element is derived by equivalent method,and the breathing effect of shaft crack is simulated by cosine function model.After that,the modal analysis and response solution of the system are carried out,and the fault characteristics are obtained by comparison.At the same time,the maximum correlation kurtosis deconvolution algorithm(MCKD),variational mode method(VMD)and fast spectral kurtosis method(FSK)are combined,and particle swarm optimization(PSO)algorithm is used to select the parameters of MCKD and VMD algorithm,so as to process the experimental data and verify the correctness of the dynamic model.The results show that:(1)Affected by the shaft crack,the overall stiffness of the system decreases,so the natural frequency of the system presents an overall downward trend,and with the increase of the degree of damage,the natural frequency decreases more obviously;at the same time,there is an obvious "inflection point" in the vibration mode diagram at the crack location.(2)Under the influence of crack breathing,the time-domain response appears obvious "simple harmonic" modulation,and the modulation period is the reciprocal of the fault shaft rotation frequency;in the low frequency region of frequency domain,the rotation frequency and its frequency doubling components appear,and a small number of sidebands appear near the meshing frequency and frequency doubling,and the sideband interval is the fault shaft rotation frequency.(3)Through comparative analysis,it is concluded that the inflection point of the system vibration mode can be used to identify the damage location.However,with the increase of fault degree,the lower the natural frequency is,the more serious the time-frequency modulation phenomenon is,and the amplitude range of signal fluctuation increases,so that the peak to peak value increases with the deepening of damage degree.The RMS and kurtosis factor decrease with the deterioration of the damage,and the kurtosis increases first and then decreases.(4)MCKD-VMD-FSK algorithm can effectively extract the features of the transmission system signal with shaft crack,and the algorithm has good robustness,and the correctness of the simulation model is verified. |