K-NEAREST NEIGHBOR CLASSIFIER (KNN) FOR STATOR WINDING SLOT FORM IN INDUCTION MOTORS


Creative Commons License

Noğay H. S.

6. ULUSLARARASI MÜHENDİSLİK, MİMARLIK VE TASARIM KONGRESİ, İstanbul, Turkey, 17 - 18 December 2020, pp.675-684

  • Publication Type: Conference Paper / Full Text
  • City: İstanbul
  • Country: Turkey
  • Page Numbers: pp.675-684

Abstract

Abstract: Six types of stator winding slot forms can be used in rotating electric machines. The operation of rotating electrical machines with minimum losses within the desired load limits is directly related to the form of the slot where the stator windings are placed, like all other components in the machine. For each stator winding slot form, the groove leakage flux, the tooth head leakage fluxes depending on the slot form, the coil head leakage flux differ in characteristic terms at different loads and different stator winding structures. In the machine design process, the stator winding slot forms are automatically decided by looking at the preferred power values of the machine to be designed. However, for designers, considering the flute leakage in deciding the stator winding slot form can provide an important advantage in terms of the losses caused by the slot leakage fluxes and therefore the machine's performance. Slot form is normally considered standard for three-phase cage induction motors. Therefore, the stator and rotor winding slot forms are not parameters that can be changed immediately by the designers in order to obtain the desired outputs, taking into account the operating conditions of the motor. Including the groove shape among the variable parameters during the design phase, the number of components that can be changed for designers will increase, and more advantageous (such as silent running, etc.) advanced motors can be designed. In this study, stator winding groove shape classification has been performed using the “k-nearest neighbor classifier”, which is one of the populer machine learning methods. From the results, it is understood that using machine learning methods in deciding the slot form can save a significant amount of time in machine design and lead to more advanced quality machine designs.