Artificial Neural Network Based Designing System for Induction Motors

Abstract

An artificial neural network (ANN) based designing method has been proposed for induction motors. Based on the real design data for various types of induction motors with different rating, three-layered cascaded artificial neural networks have been specified and trained to estimate the fundamental quantities and parameters, and also to give proper sizes of the target induction motor. Totally, thirty-four cage rotor type induction motors, which were designed by different engineers, have been selected for the training data. These induction motors have different ratings, different capacities, and different usage, i. e., driving units for pumps, compressors, and fans. The type of cooling method is totally enclosed the fan cooled type for all the selected induction motors. Through the numerical simulations based on the actual design data for different types of induction motors, the efficiency of the proposed designing system has been demonstrated.