The Research On Neural Network Inverse Control For High-speed Motorized Spindles And Simulation

The high-speed motorized spindle is the key components to incarnate theperformance of high-speed machine tools, which combines the motor rotor with thespindle. So its drive control performance has a significant impact on the machine tool’smachining quality. The V/F control is a kind of steady-state control method, therefore itsdynamic performance is not stable. By using of coordinate transformation, the vectorcontrol transforms the multi-variable and strong-coupling motor model into a simplemodel, which is similar to DC motor. In the view of the control method, the vectorcontrol just achieve approximate the steady-state decoupling control, and it over-dependon the system parameters. In the view of this, the paper will elaborate on the method ofcombining the inverse system and neural network to linearize and decouple the complexnonlinear systems, like the high-speed motorized spindle.Firstly, the background and significance of this research is discussed, the researchstatus and development trend of high-speed spindle drive control are summarized indetail, and the fundamental structure of high-speed motorized spindles is introduced andits operation principle is analyzed. And the mathematical model of the motorizedspindle is derived, which shows that the motorized spindle is a high-order,multi-variable, non-linear and strongly coupled electromechanical system.Secondly, the principles of inverse system linearization decoupling and neuralnetwork are introduced. The superiority of the neural network inverse control strategy isexpounded. According to the state equation of the high-speed electric spindle， thereversibility of the spindle system is proved.Thirdly, the process of achieving the neural network inverse control strategy isproposed, which includes confirming the structure of the neural network inverse control,constructing complex controller based on the neural network inverse control. Blockdiagram of neural network inverse control on the high-speed spindle system isconstructed. A simulation model of the spindle’s neural network inverse control systemis established in Matlab/Simulink.Finally, the parameters of the type170MD15Y20high-speed motorized spindle areused in the model. The neural network inverse control are compared with the vectorcontrol.In this way, the feasibility and superiority of the neural network inverse control strategy is proved.