Parameter Estimation of Squirrel-Cage Induction Motors Using the Optimization-Powered XGBoost Algorithm

Induction machines have a broad array of applications in various fields, from transport to healthcare. It is necessary to know the equivalent circuit parameters to determine the behavior of these machines under different load conditions and to control the speed position. Equivalent circuit parameters can be obtained experimentally using the IEEE 112 standard. In addition, the artificial intelligence models demonstrate a capacity for highly precise estimation of the parameters based on the data obtained from the machine manufacturers instead of using approximation formulas. In the study, the equivalent circuit parameters of squirrel-cage induction motors with different power ratings were estimated for single and double-cage models using the XGBoost algorithm with metaheuristic optimization. The proposed method has been scrutinized with the existing studies for both models and its performance has been studied. With the proposed algorithm, more successful results were obtained for the single-cage model than other algorithms. Furthermore, the results obtained for the double-cage model are competitive with other methods. In this study, the Particle Swarm optimization algorithm, a well-known metaheuristic optimization algorithm, is applied to determine the optimal configuration of the Extreme Gradient Boosting (XGBoost) algorithm to improve the prediction accuracy of the Machine Learning (ML) model. The results clearly show that the proposed approach is better than the existing literature. The performances of the proposed approach on the single and double cages are obtained with Mean Absolute Percentage Error (MAPE) values of 1.79% and 6.36%, respectively.