Since the shape of the groove in which the stator winding is placed in the rotating electric machines changes the air gap inside and outside the groove outside the coil, it affects all other factors, including leakages and losses, which affect the performance of the electric motor. How different groove shapes affect motor losses and magnetic leakages can be considered as one of the subjects to be investigated. In electrical machines, leakage reactances are usually not taken into account since the designers decide the groove shape automatically via packaged programs. During the design, the drawings of different groove types can be of different sizes according to the dimensions of the groove. It can be very beneficial for the performance of the electric motor to draw the groove drawing shapes by considering the groove leaks or to decide on the most suitable one among the previously drawn groove shapes. In this study, the convolutional neural network model, which has been the most popular among deep learning methods especially in recent years, has been used in order to help decide on the stator groove shape. According to the results of multiple classification made with the help of CNN, it was seen that very successful results were obtained in deciding the shape of the groove. It has been observed that deep learning can facilitate and speed up the work of designers in deciding or estimating the groove shape in electric motor design.
Anahtar Kelimeler: CNN, stator groove shape , transfer learning, classification
|
