Dynamic Equivalent Modeling of Induction Motors based on K-means Clustering Algorithm

Abstract

It is unrealistic to establish the model for analyzing dynamic characteristic of each induction motor. In order to ensure that the equivalent induction motor models impose similar impact on the electric system compared with the original induction motors group, the critical slip scr and the load rate KL of the induction motor were taken as the clustering indicators in this paper. Induction motors group was clustered based on the K-means algorithm to get different cluster numbers. Then the optimal clustering result was obtained by the F statistics. After the clustering, induction motors in different classes were equalized and the accurate dynamic equivalent models were established. The simulation curves were obtained after comparing the established equivalent models with the original models and the classic models in the short circuit fault condition through the example system. The simulation results proved that the method proposed in this paper were reliable.