Speed control of permanent magnet synchronous motor using artificial neural network

The performance analysis on the principle of operation, design considerations and control algorithm of the field oriented control (FOC) for a permanent magnet synchronous motor (PMSM) drive system in close loop operation are presented in this study. To improve the speed tracking and load disturbance performance of PMSM drive system, an online learning backpropagation algorithm Artificial Neural Network (ANN) speed controller by feedforward architecture is proposed. The proposed controller was compared to offline learning ANN in order to show the performance differentiation between both online and offline learning methods. The ANN for both speed controllers are defined as 1-3-1 network structure. Using the output of the ANN speed controller, the quadrature axis reference current value can be obtained. While, for the direct axis reference current value is set to zero. Then, the value of direct and quadrature axis current are produced by FOC in order to decouple the torque and flux. These values differentiations are become the input of torque and flux controller that used proportional integral (PI) controller. The capability of the proposed ANN has been verified in simulation model by using Simulink/Matlab and experimentally by using Digital Signal Processor Controller. From the simulation and experimental results show that the proposed online learning ANN speed controller adaptively tackles the load variations and enables the drive system to follow the reference speed quickly compared to offline learning ANN speed controller. It can conclude that, the proposed online learning ANN speed controller has better performance rather than offline learning ANN speed controller in term of settling time, speed error after steady at one point and momentary under speed