Alt birimler

Araştırmacılar(145)

Birim yayınları

949 →
  • ArticleYalnız künye2027
    Single-image generation using multi-scale vector quantization with autoregressive prior

    Yildiz, Eyyup

    Single-image generative models aim to learn internal statistics of a single training image and synthesize diverse, realistic samples without using large-scale datasets. However, current models rely either on multi-scale generative adversarial network (GAN) pyramids or on diffusion models, both of which require controlling the continuous latent space of the training image. In this paper, we introduce a new single-image generative model that utilizes image-specific discrete latent representation by integrating vector quantization with a coarse-to-fine multi-scale pyramid. Vector quantization serves as a dictionary of high-frequency textural features, ensuring preservation of local textures and fine details that are typically lost in continuous latent space. To generate new image samples, our model trains an autoregressive prior over the quantized token maps, allowing image synthesis to be performed in a quantized latent space rather than directly in pixel space. Extensive quantitative and qualitative experiments on Places50, SIGD, and COCO50 datasets demonstrate that our model consistently achieves competitive performance in image quality, diversity, and computational efficiency, as measured by SIFID, NIQE, NIMA, LPIPS, and MS-SSIM, compared with existing models. We further demonstrate the versatility of our model through various image manipulation tasks, including paint-to-image, harmonization, editing, and outpainting. In conclusion, we show that vector-quantized latent modeling is an effective alternative to GAN- and diffusion-based approaches for high-quality and diverse single-image generation and represents a notable improvement over existing models.

    RDL Kalıcı Kaydıİncele
  • ArticleYayıncıda açık erişim2026
    Machine Learning-Based Modeling and Performance Analysis of a Novel Battery Thermal Management System with a Ternary Periodic Minimal Surface Porous Structured Liquid Cooling Channel

    Çolak, Andac Batur; Kilinç, Cuma

    This article examines the fundamental requirements for effective thermal management of batteries, particularly in high-energy density applications such as electric vehicles and energy storage systems. A unique method is introduced that utilizes a triple periodical minimal surface pore structure in liquid cooling channels to improve heat dissipation and temperature homogeneity in battery packs. Experimental validation and optimization via artificial neural networks are used to evaluate and improve thermal performance, focusing on critical metrics such as maximum temperature and temperature difference within the battery system. Artificial neural network models demonstrated model reliability, exhibiting ideal prediction accuracy with low mean square error values of 1.29E-05 and 9.87E-06, and a correlation coefficient of 0.99999. AThis work builds upon previous attempts by combining an innovative structural design with machine learning, aiming to provide a robust and data-driven solution that aligns cooling efficiency with structural integrity. The novelty of the study lies in its dual experimental machine learning approach, which provides a foundation for improved predictive control in battery temperature management systems with high accuracy and minimal resource requirements.

    RDL Kalıcı Kaydıİncele
  • ArticleYayıncıda açık erişim2026
    Multi-Modal Attention-Based Automatic Diagnosis of Bearing Vibration Faults for Robust Mechanical Condition Monitoring

    Tastimur, Canan; Kaya, Volkan

    Bearing components are critical elements of rotating machinery, and their failures can lead to sudden shutdowns, reduced efficiency, and increased maintenance costs. To address the limitations of conventional fault diagnosis methods, this study proposes a multimodal deep learning framework that integrates raw time-domain vibration signals with stationary wavelet transform (SWT)-based time-frequency representations using a novel bidirectional cross-attention mechanism. This architecture enables effective and mutual interaction between heterogeneous features, significantly enhancing discriminative representation learning and fault classification performance in complex operating environments. The proposed model was evaluated on the public SUBVF1.0 dataset under three fault conditions, achieving training and validation accuracies of 99.17% and 95.63%, respectively. The model also demonstrated strong performance in terms of precision, recall, F1-score, and ROC analysis. An ablation study further confirms the contribution of each model component. These results indicate that the proposed approach is accurate, robust, and suitable for practical industrial fault diagnosis applications.

    RDL Kalıcı Kaydıİncele
  • ArticleYayıncıda açık erişim2026
    A New High Gain Single-Switch Quadratic Boost Converter with Low Voltage Stress

    Ting, Naim Suleyman; Ipek, Asli Seyda

    Recent progress in DC microgrids necessitates high gain DC-DC converters for effective energy conversion. In this regard, this paper introduces a new high gain single-switch quadratic boost converter (HGS2QBC). This converter, comprising a switch, two inductors, four capacitors, and five diodes, incorporates a voltage multiplier cell (VMC) that provides an 8-times voltage gain at a 0.5 duty cycle. The salient features of the proposed converter are its high voltage gain, having only one switch, low voltage stress across semiconductor devices, and non-pulsating input source current. In addition to its high voltage gain, one of the most significant added-value features of the proposed converter is the presence of a single switch in its structure and the fact that the voltage stress on this switch is half of the output voltage. Furthermore, a 100 W, 200 V, and 20 kHz laboratory prototype is constructed to assess practical viability, and the related experimental results are provided. Experimental analysis confirms the converter's effectiveness in managing voltage and current waveforms across its components, reducing stress on the semiconductor elements and minimizing voltage levels across the diodes. Such design optimizations not only provide superior performance but also offer cost advantages that make the converter economically viable.

    RDL Kalıcı Kaydıİncele
  • ArticleYalnız künye2026
    A new full insulated multi output DC-DC converter for photovoltaic energy regulation and multi-battery charging

    Can, Erol

    The integration of solar energy into electronic devices and battery-charging systems is highly valued due to its status as a clean, environmentally friendly power source. To satisfy energy demands while minimising environmental impact, it is essential to adopt efficient, sustainable technologies. This study introduces a multi-output (dual-inductor, single-capacitor) LLC DC-DC converter designed for battery charging applications, capable of independently charging multiple batteries while reducing input voltage fluctuations from photovoltaic (PV) sources. The proposed converter features independently operated coils for each output channel, preventing long-duration short circuits in the inductors. It achieves high-voltage conversion with reduced output voltage variation, offering greater safety and stability compared to conventional converter designs. A key advantage of this system is its ability to deliver different voltage levels across outputs, enabling flexible, simultaneous charging of various batteries and loads. Initially, circuit modelling and system definition are performed to illustrate these benefits. The dual-inductor, dual-output converter is then evaluated under varying temperatures and irradiance levels during the application phase. Results show reduced ripple levels and increased PV output voltage with lower oscillation while maintaining electrical isolation between batteries and loads.

    RDL Kalıcı Kaydıİncele
  • ArticleYayıncıda açık erişim2026
    LandTrendr Algoritması ve Makine Öğrenme Kullanılarak Orman Tahribatı ve İyileşme Süreçlerinin Sınıflandırılması: Milas-Bodrum Örneği

    Taş, Mehmet Akif

    Bu çalışma, zengin orman varlığına rağmen Türkiye'nin yangına en hassas bölgelerinden olan Muğla'nın Milas-Bodrum yöresindeki orman dinamiğini, 2013-2024 periyodu için uzaktan algılama ve makine öğrenmesi yöntemleriyle analiz etmektedir. Analizde, 49 adet Landsat 8 görüntüsünden oluşan zaman serisi verileri kullanılmıştır. Normalize Edilmiş Yanma Oranı (NBR) indeksine dayalı bu veriler, LandTrendr algoritması ile işlenmiş ve ardından Random Forest makine öğrenmesi modeliyle her yıl için "Sağlıklı Orman", "Bozulma" ve "İyileşme" olmak üzere, ekosistemin durumunu yansıtan üç dinamik sınıfa ayrılmıştır. Kantitatif bulgular, 2014-2021 arasında "Bozulma" oranının %24'e gerilediği bir iyileşme döneminin, 2021'deki büyük yangınla dramatik şekilde kesintiye uğradığını ortaya koymuştur. Hem sağlıklı hem de iyileşmekte olan ormanları etkileyen bu olay, "Bozulma" oranını tek bir yılda %47'ye fırlatmış; ancak 2024 itibarıyla "İyileşme" kategorisi tekrar %29'a ulaşarak yeni bir rejenerasyon döngüsünün başladığını teyit etmiştir. Bu çalışma, geleneksel sınıflandırmaların ötesine geçerek, makine öğrenmesi ve zaman serisi analizinin büyük afetlerin ekosistem etkilerini belgelemek ve iyileşme süreçlerini izlemek için sunduğu güçlü metodolojik çerçeveyi ortaya koymaktadır.

    RDL Kalıcı Kaydıİncele
  • ArticleYayıncıda açık erişim2026
    2023 Kahramanmaraş Depremleri sonrasında Malatya’daki yığma binaların sismik performansı: Saha incelemesi ve nümerik analizler

    Turan, Ahmet Ihsan; Celik, Alper

    Türkiye, aktif deprem kuşağında yer almakta olup tarih boyunca yıkıcı depremler yaşamıştır. 6 Şubat 2023 Pazarcık (Mw 7.7) ve Elbistan (Mw 7.6) depremleri, özellikle bölgede yaygın olan mühendislik hizmeti almamış yığma yapılarda büyük hasara yol açmıştır. Bu çalışma, Malatya ilindeki yığma yapı stokunda gözlenen hasar türlerini incelemekte ve saha gözlemlerine dayanarak bu hasarların olası nedenlerini tartışmaktadır. Nitel bulguları desteklemek amacıyla, temsilî az katlı yığma yapı prototiplerinin deprem davranışını karşılaştırmalı olarak değerlendiren sayısal analizler gerçekleştirilmiştir. Doğrusal-elastik yanıt spektrumu analizleri ve modal esaslı yaklaşım kullanılarak iki ana parametre dikkate alınmıştır: taşıyıcı duvar malzemesi türü (dolu kil tuğla, boşluklu kil tuğla ve gazbeton) ve kat sayısı (1–3). Saha gözlemleri, yapı stokunun önemli bir bölümünün yeterli mühendislik hizmeti olmadan inşa edildiğine işaret etmekte; ayrıca katlar arasında farklı yığma birimlerinin kullanılması gibi düzensiz uygulamaların kritik zafiyetlere yol açabildiğini göstermektedir. Sayısal sonuçlar da rijitlik ve deplasman taleplerinin hem yığma birimi türüne hem de yapı yüksekliğine güçlü biçimde bağlı olduğunu ortaya koymakta ve özellikle daha hafif yığma sistemlerde kat sayısı arttıkça daha olumsuz talep eğilimlerinin gelişebileceğine işaret etmektedir.

    RDL Kalıcı Kaydıİncele
  • ArticleYayıncıda açık erişim2026
    Görüntü Şifrelemede RC4 ve Kaotik RC4 Yöntemlerinin Karşılaştırmalı Performans Analizi

    Akar, Funda; Karhan, Muhammed Baki

    Dijital görüntülerin güvenliği hem bireysel kullanıcılar hem de kurumsal uygulamalar için büyük bir öneme sahiptir. Özellikle görüntülerin şifrelenmesi, veri güvenliğini artırmak ve yetkisiz erişimleri engellemek için etkili bir yöntem olarak kullanılmaktadır. Bu bağlamda, simetrik şifreleme algoritmaları ve kaotik sistemlerin bir arada kullanımı, son yıllarda yoğun ilgi görmüştür. Bu çalışmada, yaygın olarak kullanılan RC4 şifreleme algoritması ile kaotik bir sistem olan Logistic Map tabanlı güçlendirilmiş RC4 yöntemi kullanılarak görüntü şifreleme ve şifre çözme işlemleri gerçekleştirilmiş ve karşılaştırmalı analizler yapılmıştır. Performans değerlendirmesi, histogram analizi, korelasyon analizi ve entropi analizi gibi yöntemlerle gerçekleştirilmiştir. Elde edilen sonuçlar, RC4 algoritmasının, görüntü şifreleme ve şifre çözme süresi kaotik sistemle güçlendirilmiş RC4’e göre daha hızlı bir performans sergilediğini göstermiştir. Ancak, kaotik sistemle güçlendirilmiş RC4 yöntemi, histogram, korelasyon ve entropi analizlerinde daha başarılı bulunmuş ve bu yönüyle daha yüksek bir güvenlik sağladığı görülmüştür.

    RDL Kalıcı Kaydıİncele
  • ArticleYayıncıda açık erişim2026
    Decoding brain signals for inner speech: eeg signal sub-band decomposition and machine learning approach

    Karakas, Mehmet Fatih

    Decoding inner speech is a rapidly advancing field with significant implications for brain-computer interfaces (BCIs) and assistive technologies. In this study how sub-band decomposition, feature types, and classifier choice affect inner-speech classification were assessed. A comprehensive comparative framework is presented, employing various sub-band decomposition techniques including Tunable Q-Factor Wavelet Transform, Empirical Mode Decomposition, Variational Mode Decomposition, and Circulant Singular Spectrum Analysis. From the decomposed signals, 17 statistical, 4 entropy, and 5 frequency features were extracted individually and in hybrid combinations before reduction with Least Absolute Shrinkage and Selection Operator. Classification performance was evaluated with 10-fold cross-validation, using Support Vector Machine, Feedforward Neural Network, and K-Nearest Neighbor. Performance was summarized as the mean across 30 independent runs per condition over six metrics. A three-way repeated measures ANOVA with Greenhouse-Geisser correction showed significant main effects and interactions across all metrics (p < 0.001), with large effect sizes (η2p >= 0.14). The best overall configuration was EMD-Hybrid-SVM outperforming baseline models, yielding a group-level mean accuracy of 54.13%. These findings demonstrate the potential of sub-band decomposition and hybrid features for improving inner speech decoding and contribute to the development of more robust EEG-based BCIs.

    RDL Kalıcı Kaydıİncele
  • ArticleYayıncıda açık erişim2026
    Rate-maximized cooperative su selection based on eigenvalues for cognitive radio networks

    Ilgın, Fatih Yavuz

    It is clear that the future of wireless communication systems will face the problem of spectrum scarcity. Cognitive radio systems currently hold a significant position as a solution to this problem. Cognitive radios scan for spectrum gaps and make available for reuse of empty spectrum bands. In this study, we proposed a novel threshold for this method using a different cooperative spectrum sensing method with Standard Condition Number (SCN) for cognitive radios. Detailed simulation studies were included to evaluate the performance of the method. In the simulation studies, different eigenvalue-based sensing methods and energy-based sensing methods were compared using MIMO-OFDM based communication systems. The simulation studies proved the efficiency of the proposed cooperative method.

    RDL Kalıcı Kaydıİncele
Tam Kayıt
AlanDeğerDil
dc.date.accessioned2026-10-20T15:26:39.240248198Z
dc.date.available2026-10-20T15:26:39.240248198Z
dc.identifier.urihttps://rdlopen.rawdatalibrary.net/r/ebyu/22947
dc.rightsinfo:eu-repo/semantics/openAccess
dc.titleMühendislik ve Mimarlık Fakültesitr
dc.titleFaculty of Engineering and Architectureen
dspace.entity.typeOrgUnit
organization.identifier.codemuhendislik-mimarlik
organization.unitTypefaculty
rdl.unit.alias\bengn (&|and) archit
rdl.unit.aliasengineering (and|&) architecture
rdl.unit.alias\bfac engn\b
rdl.unit.aliasfaculty of engineering
rdl.unit.aliasmuhendislik
rdl.unit.aliasengineering faculty