Research On Active Vibration Control Of Gear Transmission System Based On Neural Network

As an important part of mechanical transmission equipment,gearboxes are widely used in aerospace,vehicle transportation,lathe processing and many other major engineering fields.The gear vibration problem widely occurs by the transmission error excitation,and radiate noise to the box through the bearing.In serve cases,the gear vibration will lead to the instability of the transmission system,reduce the performance and life of the transmission system.Therefore,reducing the vibration of the gear system is of great engineering significance for reducing the gearbox fault and improving the working environment.Aiming at the periodic vibration caused by meshing error in gear transmission system,according to the idea of active vibration control,two-stage gear transmission system is taken as the research object and the installation position of the actuator on the shaft is optimized by particle swarm optimization algorithm.Meanwhile,the BP neural network is proposed as the control algorithm,combined with the online identification to obtain the secondary channel model,to control the gear in meshing nonlinear vibration at fundamental frequency and higher harmonic.The specific research contents are as follows:(1)Based on the multiple-degree-of-freedom dynamic finite model of one two-stage gearbox,which is established by the Dynamic Substructure method,and the mechanical-voltage coupling model of the piezoelectric material actuator,the position of the actuator on the shaft in the gearbox is determined by using the system controllability criteria and the particle swarm algorithm.(2)In order to effectively suppress the periodic vibration in the gear transmission system,the model reference indirect adaptive control structure based on neural network is adopted.On the premise of system stability,the internal parameters that affect the neural network algorithm are analyzed,and the influence of secondary channel identification error on the performance of active control strategy is analyzed.Aiming at the slow convergence speed of neural network algorithm,momentumfactor and correction factor are introduced to improve the weight updating algorithm of neural network algorithm.(3)The gear transmission system vibration mathematical model is combined with neural network algorithm to carry out simulation experiments.To reduce the influence of the reference signal error on the stability and convergence of the active vibration control system,vibration error signal is introduced as an input of neural network algorithm to improve the control strategy structure.The reference signal amplitude ratio has a greater impact on the control effect.(4)Performing simultaneous multi-frequency harmonic control experiments under the condition of different speeds,there is a 3dB-10 dB vibration reduction for each target frequency after adjusting the amplitude ratio of the reference signal,proving the effectiveness of the proposed active control strategy.