Demagnetization fault detection of permanent magnet synchronous motor with convolutional neural network
Yazarlar (2)
Tokat Gaziosmanpaşa Üniversitesi, Türkiye
Doç. Dr. Mustafa EKER
Tokat Gaziosmanpaşa Üniversitesi, Türkiye
| Makale Türü | Özgün Makale |
| Makale Alt Türü | SSCI, AHCI, SCI, SCI-Exp dergilerinde yayınlanan tam makale |
| Dergi Adı | Electrical Engineering |
| Dergi ISSN | 0948-7921 Wos Dergi Scopus Dergi |
| Dergi Tarandığı Indeksler | SCI-Expanded |
| Dergi Grubu | Q3 |
| Makale Dili | İngilizce |
| Basım Tarihi | 02-2023 |
| Cilt No | 105 |
| Sayı | 3 |
| Sayfalar | 1695 / 1708 |
| DOI Numarası | 10.1007/s00202-023-01768-9 |
| Makale Linki | http://dx.doi.org/10.1007/s00202-023-01768-9 |
| Özet |
| In this study, the convolutional neural network (CNN) architecture of deep learning was used to diagnose a demagnetization fault that occurred in permanent magnet synchronous motors (PMSM) under stationary speed conditions. Faults in the motor increase the maintenance costs and decrease the production capacity. At first, it was tried to prevent faults with periodic maintenance, but it was not enough to prevent disruptions in production because of faults. For this reason, condition monitoring methods in electric motors have recently replaced periodic maintenance. PMSMs are widely used in industry due to their advantages, such as high efficiency and high power-to-weight ratio. Nowadays, the CNN, one of the deep learning architectures, is widely used due to its success in classification. In the study, a new CNN architecture was created for the detection of demagnetization faults in PMSM. The signals used in the CNN architecture in this study are the current signals of the PMSM. As a result of the study, the proposed CNN model achieved a high success rate of 99.92% on average in diagnosing the demagnetization fault in PMSM. |
| Anahtar Kelimeler |
| CNN | Deep learning | Fault detection | Demagnetization fault | PMSM |
BM Sürdürülebilir Kalkınma Amaçları
| Atıf Sayıları | |
| WoS | 20 |
| SCOPUS | 21 |
| Google Scholar | 34 |