Effective identification of induction motor parameters based on fewer measurements

This paper applies genetic algorithms (GAs) to the problem of parameter identification for field orientation control (FOC) induction motors. Kron's two-axis dynamic model in per-unit system is given, and the model's parameters are estimated by a GA using the motor's dynamic response to a direct on line start. Results with different levels of measurement noise are presented for the model both in the per-unit system and in actual values. For comparison, the results of a simple random search (SRS) method under the same condition are also given. The results show that the parameter identification accuracy, the convergence speed and the practicality of the algorithm have been improved significantly by use of the model in the per-unit system. The results also show that fewer measurements are required to identify the induction motor parameters accurately.