Akademik Veri Yönetim Sistemi
Tezin Türü: Doktora
Tezin Yürütüldüğü Kurum: Yıldız Teknik Üniversitesi, Elektrik-Elektronik Fakültesi, Elektronik Ve Hab.Müh.Böl, Türkiye
Tezin Onay Tarihi: 2024
Tezin Dili: İngilizce
Öğrenci: Hatice Vildan DÜDÜKÇÜ
Asıl Danışman (Eş Danışmanlı Tezler İçin): Nihan Kahraman
Eş Danışman: Murat Taşkıran
Özet:
This study offers a novel approach that is well-suited for the utilization of onboard systems in UAVs for detecting anomalies in flight sensor data. To achieve this objective, the data collected from UAV sensors was analyzed, and a classification method was developed to identify abnormal flying patterns. Additionally, a regression method was also proposed to detect anomalies in UAV sensors. The effectiveness of these methods was assessed by utilizing both existing UAV datasets from the literature and a dataset specifically generated for the study. Furthermore, as part of the study, data obtained via flight sensors has been collected and labeled. Subsequently, this dataset was utilized to assess both classification and regression techniques and it served as a way to analyze the proposed hybrid approach that combines a classification method for detecting flight anomalies with a regression method for sensor analysis. In order to assess the effectiveness of the hybrid method on power-constrained devices with limited memory resources, a Raspberry Pi development board was utilized as the testing platform. In addition, an model update feature, which made it possible to conduct continuous model training on the edge, was developed and evaluated.