Akademik Veri Yönetim Sistemi
Tezin Türü: Yüksek Lisans
Tezin Yürütüldüğü Kurum: Dokuz Eylül Üniversitesi, Mühendislik Fakültesi, Elektrik - Elektronik Mühendisliği Bölümü, Türkiye
Tezin Dili: İngilizce
Öğrenci: Gürcü Nur Ayas
Danışman: Taner Göktaş
Özet:
In order to avoid costly shutdowns, minimize safety concerns, and obtain high performance over long periods of time, electrical motors used in industries should be monitored carefully. The existing commercial monitoring solutions for electrical machines can sometimes mostly bulky, costly, and may not be practical to use in various applications. Among the diagnosis and prognosis methods, the phase current information is mostly preferred in electrical motors since it is readily available in motor drives for control purposes. However, studies have shown some major shortcomings of current-based monitoring. For instance, the fault signatures in the current spectrum depend on stator winding configuration, winding type, fault location, controller gains, load level, and so on. Therefore, setting a threshold without positive or negative false alarms is a challenging task. As an alternative, stray flux, vibration and acoustics spectra can be used to any electrical and mechanical faults, such as broken bar, eccentricity, short circuit, misalignment etc., as it facilitates reliable diagnosis and enables remote or on-case monitoring options through small foot-print integrated sensors. In this study, a reliable detection of electrical and mechanical faults through multi-sensor data fusion can be provided in electrical machines. This approach minimizes the additional false positive signatures in corresponding spectra and gives more accurate results for condition monitoring process. This study can provide an effective solution in terms of both academia and industrial applications.