A new neural network based approach for speed control of PM synchronous motor

In this paper a new Artificial Neural Networks (ANN) speed tracking control design for Permanent Magnet synchronous Motors (PMSM) is presented. The traditional Proportional-Integral (PI) regulator is widely used in vector Control scheme of the synchronous motor because of the robustness this regulator procures. Even so, the performances derived by these linear controllers are usually limited due to their sensitivity to non linear behaviour of PMSM dynamics, load disturbance, and parameter uncertainties. The Artificial Neural Networks seems to be a promising solution in this purpose. In this work, a neural network controller is developed, for speed control of high performance PMSM drive systems, to improve efficiency of a vector control drive. Because of the electrical model of motor is nonlinear two inputs two outputs systems which need a multivariable current controller, we propose in the second parts, a new multi-input multi-output (MIMO) neural network controller to replace both of the speed regulator and d-q axis currents regulators. The proposed controllers, procures good transient performance, load rejection and robustness against parametric uncertainties. Moreover the ANN controllers provide high steady state performances. Simulation results are presented to illustrate the performance of the proposed controllers under the various conditions.