Rotary electric machines are widely used in every field of industry. The performance of an electrical machine during its operation in the industry directly affects not only its efficiency of itself but also the efficiency and production capacity of the workplace. Therefore, the design of electrical machines is a process that affects the important parameters of the machine in terms of both meeting the needs of the industry and achieving the most optimal results with the least loss. Nowadays, package software programs are used in the design process of rotating electrical machines. However, in these design programs, the stator winding groove shapes are automatically selected. However, considering the leakage fluxes in the design, it is necessary to decide on the shape and size of the groove. Losses caused by leakage fluxes are one of the factors affecting motor efficiency and power. The shapes and dimensions of the grooves in which the stator windings are placed in rotating electrical machines directly affect the groove leakage fluxes, tooth head leakage fluxes and losses caused by other leakages. Because when placing the stator windings in the grooves, it is not taken into account where the coil edge is placed in the groove, and each groove has a different air gap. For this reason, even if the winding type, groove shape, coil pitch, coil length, number of poles and cross section of the coil are exactly the same, power differences may occur in electrical machines. Therefore, having the same groove shape can also prevent power differences. Deciding on the shape of the groove at the design stage, taking into account the groove leakage field, can provide a significant advantage for designers. Performing this process by making use of machine learning methods is an important factor that accelerates the design process. In this study, support vector machines (SVM), which are still popular among machine learning methods, are used in determining and classifying the groove shape of rotating electrical machines by taking into account the groove leakage. In the study, it has been concluded that the SVM method can make a great contribution to the stator groove shapes of the rotary electric machines in the decision-making process in the design process.
Anahtar Kelimeler: SVM, groove leakage, classification, machine learning