Neural Network Inverse Based Decoupling Control For Pmsm Drive System

Permanent magnet synchronous motor(PMSM)drive system has the features such as strong coupling,nonlinear and multivariable.The traditional control method can only achieve approximate steady-state decoupling control,and is sensitive to external disturbances,motor parameter changes.Its robustness is poor and the control performance is not ideal.Neural network has strong robustness,adaptability,approximation ability and nonlinear mapping ability,and can achieve very accurate approximation of nonlinear system.In this paper,a neural network inverse decoupling control method is proposed based on the combination of neural network and inverse system method.The permanent magnet synchronous motor system is studied and analyzed in the full speed range.Based on the mathematical model of permanent magnet synchronous motor(PMSM),the basic principle of the linearized decoupling control of the inverse system is described,and the reversibility of the system is proved.Aiming at linear resolve decoupling control strategy of poor anti-interference ability poor robustness and parameter faults,neural network inverse decoupling control strategy is proposed.The additional rotational speed and the additional current controller are designed based on the sliding mode variable structure,together with the neural network inverse decoupling controller for permanent magnet synchronous motor closed-loop control.This method avoids the disadvantages such as sensitivity to motor parameters and external disturbance,poor robustness of parameters and so on.The method has excellent anti disturbance ability and good parameter robustness.According to the results of theoretical analysis,the control experiment platform of PMSM is built.The neural network inverse decoupling control strategy based on the compound control is applied to the permanent magnet synchronous motor(PMSM)system under two conditions: above the base speed and below the base speed.The experimental results show that the neural network inverse based decoupling control strategy can obtain good dynamic and static performance in full speed range.