NEURAL NETWORK ESTIMATION OF MUTUAL INDUCTANCE VARIATION FOR A SHADED-POLE INDUCTION MOTOR

Authors : Emre ÇELİK

Pages : 36-45

View : 23 | Download : 8

Publication Date : 2019-01-01

Article Type : Research Paper

Abstract :Shaded-pole induction motors insert ignore into journalissuearticles values(SPIMs); are often preferred in small power applications owing to their ability to work with single-phase power source, simple structure and low-cost properties. Such motors are within the class easy to manufacture, but the most difficult to analyze mathematically due to the fact that they have a variable air gap and elliptical rotating magnetic field, which leads to highly complex inductance calculations. Considering that the identification accuracy of phase variables is directly related to the correct knowledge of inductances in AC machines, the authors of this article attempt to realize a neural network insert ignore into journalissuearticles values(NN);-based inductance estimation in-between the stator and rotor, and also in-between the shading ring insert ignore into journalissuearticles values(shaded-pole winding); and rotor loop for an industrial SPIM. For this aim, corresponding inductance measurements are made first experimentally in terms of each 3.6º electrical position, and as such, a total of 101 data samples have been collected. %70 of them are considered as training data to train the NN while the remainder is adopted for testing the generation capability of NN. Results in comparison with the actual values have affirmed the excellence performance of the introduced NN in simultaneous estimation of the concerned two important inductances.
Keywords : Shaded pole induction motors, shaded pole, rotor loop, neural network, inductance estimation