Vector control of induction motors using neural network

ABSTRACTMaster ThesisVECTOR CONTROL OF INDUCTION MOTORS USING NEURALNETWORKŞükrü TAŞKAFAFırat UniversityGraduate School of Natural and Applied SciencesDepartment of Electronics and Computer Education2006, Page: 74The coupling effect between induction motor fluxes and currents are eliminated andthese two variables can be controlled independently by using the vector control technique. Thus,the motor torque and speed can be controlled by using the torque current component keeping theflux component constant. Due to the effects of time varying parameters and nonlinearmathematical model of the motor, a robust vector controller is required to obtain a correct fieldorientation.In this thesis, vector control of induction motors is presented using Neural Network(NN). NN inputs are selected as the motor current and the change of the current error and, NNoutput is the motor voltages. NN is trained using the currents errors. Feedforward and locallyrecurrent NN structure are used for the vector controller and their control performance arecompared. Vector control of the induction motor including the trained NN controller issimulated with MATLAB-Simulink. Simulation results showing the performance of the controlsystem are presented under the load and parameter variations.Keywords: Induction motors, artificial neural networks, vector control.