Detection of demagnetization fault in permanent magnet synchronous motor with convolutional neural network

With the developing technology, electric motors are the most used drive element in the industry. The widespread use of motors causes an increase in the number/types of malfunctions occurring in these elements. These malfunctions in the motor increase the maintenance costs and decrease the production capacity. At first, the malfunctions were tried to be minimized with periodic maintenance, but it was not enough to prevent the disruptions in production as a result of the malfunctions that occurred. Recently, condition monitoring methods in electric motors have replaced periodic maintenance. In condition monitoring methods in motor fault diagnosis; Deep learning method, which is a sub-branch of machine learning, is frequently used. The most widely used architecture of deep learning, Convolutional Neural Network (ESA), has become a prominent topic in motor fault diagnosis. Permanent Magnet Synchronous Motors (PMSM); It is widely used in industry due to its advantages such as high efficiency, high power-to-weight ratio. In the thesis study, a new approach is presented for the detection of demagnetization failure, which is the type of magnetic failure that occurs in PMSM. The approach presented for fault detection is the ESA architecture of deep learning. With this approach, very successful results have been obtained in motor fault detection. In this study, demagnetization failure was determined by using current data obtained from strong and faulty motors operated in stationary conditions. As a result of this thesis study, the proposed approach has achieved a success rate of 99.92 % in detecting the demagnetization failure in the PMSM. Keywords: PMSM, Demagnetization Failure, Deep Learning, ESA