The bearingless induction motor (BLIM) is a multi-variable, non-linear, strong coupling system. To achieve higher performance control, a novel neural network inverse system decoupling control strategy considering stator current dynamics is proposed. Taking the stator current dynamics of the torque windings into account, the state equations of the BLIM system is established first. Then, the inverse system model of the BLIM is identified by a three-layer neural network; by means of the neural network inverse system method, the BLIM system is decoupled into four independent second-order linear subsystems, include a rotor flux subsystem, a motor speed subsystem and two radial displacement component subsystems. On this basis, the neural network inverse decoupling control system is constructed, the simulation verification and analyses are performed. From the simulation results, it is clear that when the proposed decoupling control strategy is adopted, not only can the dynamic decoupling control between relevant variables be achieved, but the control system has a stronger anti-load disturbance ability, smaller overshoot and better tracking performance.

No abstract

No abstract

No abstract

No abstract

In the design process of the controller, the adaptive gain of model reference adaptive control (MRAC) often requires a tradeoff between the adaptive ability, robustness and stability of the control system. The tradeoff of adaptive gain leads to poor control performance and increase design difficulty. Aiming at this problem, the iterative learning idea is introduced into the model reference adaptive control strategy. The control parameter adaptive law based on the parameters of the previous control process is designed. For scalar systems, a new control strategy is constructed, which is the combination of MRAC and iterative learning control (ILC). The adaptive ability of the model reference adaptive controller is improved by using learning ability of ILC. An appropriate composite energy function is designed to prove the uniform convergence of the proposed control strategy and the boundedness of the control quantity. The proposed control strategy is applied to the ultrasonic motor. The effectiveness of the proposed control strategy is verified by experiments and simulations. The controller is designed by using the first-order model that is large different from the actual object. It verifies that the control strategy has strong robustness to model deviation and online time-varying characteristics.

No abstract

No abstract