| Abstract: | Slip-frequency vector control generally is used for variable speed induction motor drives due to its excellent response characteristics. However, it is sensitive to the variation of motor parameters since the flux current and slip-frequency commands are computed using the motor parameters. As a result, the system performance will be degraded if the controller parameters do not match the motor parameters. The authors propose a new approach of vector control in which quick responses and system robustness are obtained concurrently by the learning capabilities of neural network. In the newly developed method, controller parameters are learned in real time at medium and high speeds. Using these learned parameters, the low-speed performance of the controller can also be improved greatly. This paper describes the theoretical analysis, simulation and experimental results of the proposed method. |
