Decoupling control of induction motor based on neural networks inverse

Xianzhong Dai,Xinghua Zhang,Guohai Liu,Lei Zhang

2004-01-01

Abstract:A linearization and decoupling method named as artificial neural network (ANN) inverse for induction motor control is proposed. It is characterized by that the construction of the ANN inverse is independent of the motor model and parameters. Cascading the ANN inverse which consists of a static ANN and four integrators with the motor, the multivariable, nonlinear and strongly coupled system is decoupled into two independent second-order linear subsystems, or rotor speed subsystem and rotor flux one, so as to be easy to design the closed-loop linear regulator to control each of the subsystems. Simulation and primary experiment results show that the good static and dynamic decoupling performance and the strong robustness to both variation of parameters and load torque disturbance can be achieved by using the proposed method.