Identification of induction motor parameters using genetic algorithms

In this paper, we use Genetic Algorithms to estimate the parameters of an Induction Motor. The identification is performed using data obtained from a reference model that considers core losses and utilizes parameters previously determined through no-load and blocked rotor tests. Based on these results, the model parameters with adjustable values are estimated using Genetic Algorithms. The proposed optimization function for the Genetic Algorithms aims to minimize the weighted error of currents and speed, obtained between the data from the reference model and the simulation results of the model with adjustable parameters. The results show that when the induction motor starts by driving a load proportional to the square of its speed, the mean squared error in parameter estimation is less than 2.5 %, using a population of 20 individuals. These results, although preliminary, allow us to conclude that obtaining appropriate parameters for induction motors operating online is possible.