| Active magnetic bearing(AMB)is a novel non-contacting supporting form,which has a lot of advantages over conventional mechanical bearings,and has played an important role in various rotor systems.One main feature of AMB,is the capacity of active control,which can acheive rotor motion control.Normal rotor motion control is to identify and suppress the unbalance vibration of rotor system,for achieving higher operational precision.Nevertheless,the intricate characteristics of AMB-rotor system,such as multi-frequency vibration and broadband response,put forward high requirements to the control system.Therefore,deep research of AMB’s unbalance vibratioin control methods,to achieve suppression of multi-frequency vibration and to improve frequency-domain stability,has great significance.Rotor trajectory control is another form of rotor motion control.By fully utilizing AMB’s features of active control,rotor three-dimensional trajectory control can be implemented,to achieve additional functions such as complex surface processing and spindle attitude control.This control scheme has a bright prospect and urgently needs deep study.This paper mainly conducts research of AMB control schemes for rotor motion control,to achieve multi-frequency unbalance vibration suppression and time-frequency simultaneous control of AMB-rotor system,and to develop AMB control system of 5 degree-of-freedom(DOF)rotor for large-motion rotor three-dimensional trajectory control.Aiming at the large-motion rotor trajectory control,the integrated model of 5-DOF AMB-rotor system is established.Considering the influence of rotor motion,the AMB model is established,which includes 3 main parts: an electromagnet model with nonlinear electromagnetic force function,a circuit model of AMB power amplifier with rotor motion terms,and an integration model of signal devices such as sensors.For the needs of rotor three-dimensional trajectory control,rotor motion of all DOFs are considered.Based on the conventional 4-DOF rotor model,a 5-DOF coupling model of AMB-rigid rotor system,which includes the DOF of rotor axial translation,is established.Based on the minimum displacement criterion of unbalance vibration control,unbalance vibration control schemes of AMB-rotor system is studied.A deep neural network unbalance controller is designed and parallelly added to the PID main control.Simulation and experiment results indicate that the proposed unbalance controller reduce the unbalance vibration significantly,with a better effect than common controllers based on LMS algorithm and RBF neural network.Considering the muti-frequency and wideband fluctuating characteristics of AMB-rotor system,the time-frequency control method of AMB is studied based on the wavelet theory and time-frequency control theory,to achieve the frequency-domain control and to improve the stability of system.A novel time-frequency controller is proposed combining wavelet transform and deep neural network,which can implement the adaptive vibration control with separate multiple scales in time-frequency domain.The time-frequency analysis theory and instantaneous frequency is used in simulation and experiment analysis,and results indicate that the proposed controller implements the time-frequency simultaneous control of AMB system.Compared with normal unbalance controllers based on LMS algorithm and RBF neural network,the proposed time-frequency controller not only has the smallest vibration amplitude,but also has the narrowest bandwidth of each vibration frequency,meaning that better control of frequency-domain stability is achieved.For the trajectory control of rotor dynamic motion,nonlinear control method of AMB-rotor system is studied.Considering the highly nonlinear characteristics when rotor motion is large,sliding mode control strategy is applied as the main control algorithm,and the deep learning theory is studied to construct a deep convolutional neural network for system identification and compensation control.The novel deep convolutional neural network-sliding mode control(DCNN-SMC)algorithm is proposed.Its structure and algorithm are designed,and its stability is proofed by Lyapunov stability theory.Simulation and experiment of 2 rotor three-dimensional trajectories on a 5-DOF rotor system are implemented.The results indicate that the proposed control algorithm can well achieve the large-motion rotor three-dimensional trajectory control,and that the DCNN effectively identifies and compensates the unknown characteristics,leading to the significant increase of control precision.The proposed algorithm has obvious advantage on control precision over common RBFNN-SMC algorithm.Aiming at the large-motion rotor three-dimensional trajectory control,the dual-loop AMB control system of AMB-rotor system is designed.In the structure design of the control system,for solving the problem that rotor large motion has a strong impact on current response,a current control loop is added to the common displacement control loop,and the circuit model with rotor motion taken into consideration as well as the nonlinear electromagnetic force model are applied.In the algorithm design of the control system,sliding mode control is adopted,and the proposed DCNN-SMC algorithm is applied in the displacement control loop.Simulation and experiment of 3 rotor three-dimensional trajectories on a 5-DOF rotor system are implemented and comparison with common control schemes is analyzed.The results indicate that the proposed control system can well achieve large-motion rotor three-dimensional trajectory control.The structure of dual loop effectively eliminate the influence of rotor motion on control precision,leading to much smaller errors than common single-loop displacement control schemes.Moreover,its advantage increases when the rotor motion goes larger.In addition,the DCNN is proofed to be excellent in system identification and compensation control again,thus achieving a further improvement on control precision. |
PDF Full Text Request