PARAMETER IDENTIFICATION USING NEURAL NETWORKS IN A VECTOR - CONTROLLED INDUCTION MOTOR DRIVE

摘要

Field-oriented techniques utilizing microprocessors are now widely used for the control of induction motor drives in high-performance applications. Two generic types of field-oriented control, direct and indirect, are available to compute the rotor flux. The indirect scheme, because of its simple sensing technique is very used in the industry today; however, its performance strongly depends on motor parameters, particulary on the rotor time constant. The magnitude of the rotor resistance changes over a wide range depending on temperature, thus making it necesary an on-line tracking of its actual value. In this work, it is proposed an on-line identification of an induction motor drive's rotor time constant using neural networks and correlation techniques.