Context—Card fraud cost the global financial system over $28 billion in 2019, and losses have risen every year since. Banks hold the transaction data needed for collaborative fraud detection, but privacy regulations such as GDPR and KVKK prevent cross-institutional data sharing. Federated Learning keeps raw data local: each institution trains on its own records and sends only model parameters to a shared coordinator. Differentially-Private Stochastic Gradient Descent (DP-SGD) counters gradient inversion attacks by clipping per-sample gradients and adding calibrated noise before parameters leave the client, yielding a record-level (ε, δ) guarantee whose strength depends on the accumulated privacy budget ε. How the aggregation strategy behaves under this noise has not yet been studied.Objective—We compare five aggregation strategies—FedAvg, FedProx, cosine-similarity aggregation, FedAvg-DWA, and FedAdam—across two real-world benchmarks, two heterogeneity levels, four DP-SGD noise multipliers, and 10 random seeds per configuration. We ask whether the relative ranking of strategies survives DP noise, and what DP actually costs in deployed model behavior.Method—510 federated training runs cover five aggregation configurations, two datasets (Kaggle ULB: n = 284807, 0.17% positive rate; IEEE-CIS: n = 590540, 3.50%), two Dirichlet non-IID levels (α ∈ {0.5, 0.1}), and σ ∈ {0.5, 1.0, 1.5, 2.0}. The base learner is a three-layer MLP with LayerNorm. Privacy accounting uses the Opacus RDP accountant at δ = 10⁻⁵. Performance is reported at both the fixed 0.5 threshold and a validation-tuned F₁-maximizing threshold. All pairwise comparisons carry Bonferroni-, Holm-, and Benjamini–Hochberg-corrected p-values, and all 200 realized client-shard partitions are characterized directly.Results—On Kaggle ULB under IID partitioning, federated F₁@best-t (0.801–0.804) matches the centralized baseline (0.799 ± 0.036). Under mild non-IID without DP (α = 0.5), FedAvg-DWA narrowly leads (< 0.015 on F₁@best-t), though neither test survives correction. In the ULB privacy sweep, 7 of 40 Wilcoxon tests show nominal significance but none survive family-wise correction; performance consistently orders with FedAvg-DWA leading and FedAdam trailing. On IEEE-CIS, this ordering amplifies, and the extreme pair remains Holm-significant under DP (adjusted p = 0.020). While AUC holds near 0.95, calibration drift collapses precision at the fixed 0.5 threshold; a validation-tuned threshold recovers F₁ loss without privacy cost. Both calibration effects replicate on IEEE-CIS (ɛ = 8.10 ± 5.30 at σ = 1.0).Conclusion—Aggregation-rule differences under DP-SGD are small and dataset-dependent in magnitude, consistent in direction, and on the ULB sweep none survives correction; outcome differences arise mainly from post-training threshold tuning. We also uncover a previously unreported interaction: Opacus's DP step quietly overrides the standard loss-augmented FedProx setup, making FedProx numerically equivalent to FedAvg, verified by bit-identical per-round global weight trajectories across seeds.
Arka Plan—Bu çalışma, net sıfır emisyon hedeflerine ulaşmak için imalat sektöründe uygulanabilecek enerji dönüşüm stratejilerinin önceliklendirilmesini stratejik bir karar problemi olarak ele almaktadır. Artan sanayileşme ve ekonomik büyüme, enerji tüketimi ve karbon emisyonlarını önemli ölçüde artırmış; bu durum, imalat sektörünü net sıfır hedeflerine ulaşmada öncelikli bir alan hâline getirmiştir. Bu bağlamda, enerji dönüşüm stratejilerinin doğru şekilde belirlenmesi, çevresel sürdürülebilirlik ve ekonomik rekabet gücü açısından büyük önem taşımaktadır. Ancak enerji dönüşümü süreci; ekonomik, teknik, çevresel ve düzenleyici kriterlerin bir arada değerlendirildiği, çok boyutlu ve çoğu zaman çelişkili karar problemleri içermektedir. Mevcut çalışmalar genellikle bu kriterleri bütüncül bir çerçevede ele almakta yetersiz kalmakta ve stratejilerin sistematik biçimde önceliklendirilmesine yönelik sınırlı yaklaşımlar sunmaktadır. Bu nedenle, imalat sektöründe net sıfır hedeflerine yönelik enerji dönüşüm stratejilerinin bütünleşik ve yapılandırılmış bir karar verme yaklaşımıyla değerlendirilmesine ihtiyaç duyulmaktadır.Amaç—Bu bağlamda çalışma, enerji dönüşümüne yönelik kararların sistematik ve analitik yöntemlerle desteklenmesi gerekliliğine odaklanmaktadır. Literatürde çoğu çalışmanın “hangi teknoloji?” sorusuna yoğunlaştığı gözlemlenirken, bu çalışma “hangi stratejik dönüşüm yolu?” sorusuna yanıt vermeyi hedeflemektedir. Çalışma, enerji dönüşümünü sadece belirli teknolojilerin seçimiyle sınırlı ele almak yerine, imalat işletmeleri için uygulanabilir ve önceliklendirilebilir stratejik dönüşüm yollarını bütünleşik bir karar modeli aracılığıyla sunmaktadır.Yöntem—İfade edilen amacı gerçekleştirmek için kriter ağırlıklarının belirlenmesinde Logaritmik Yüzde Değişime Dayalı Nesnel Ağırlıklandırma (LOPCOW) ve Kriterler Arası Korelasyona Dayalı Kriter Önemlendirme (CRITIC) yöntemleri, enerji dönüşüm alternatiflerinin sıralanmasında ise İdeal Çözüme Olan Uzaklığa Dayalı Uzlaşık Alternatif Sıralama (CRADIS) yöntemi kullanılarak net sıfır bağlamında stratejik bir yol haritası çıkarılmıştır.Bulgular—Elde edilen bulgular, stratejilerin performansının S4 > S1 > S5 > S3 > S2 şeklinde sıralandığını göstermiştir. S4 stratejisi, diğer tüm stratejilere kıyasla belirgin biçimde daha yüksek performans sergileyerek anlamlı bir üstünlük sağlamıştır. S1 ve S5 orta düzeyde benzer sonuçlar üretirken, S2 stratejisi en düşük performansı göstermiştir.Sonuç—Sonuçlar, dijitalleşme ve enerji yönetimi stratejisinin net sıfır hedeflerine ulaşmada daha yüksek etkinlik sağladığını göstermektedir. Bu durum, enerji dönüşüm süreçlerinde teknolojik yatırımların yanı sıra enerji kullanımının izlenmesi, analiz edilmesi ve etkin şekilde yönetilmesinin önemini ortaya koymaktadır. Elde edilen bulgular, politika yapıcılar ve sektör paydaşları için karar destek aracı niteliği taşımakta olup, sürdürülebilir enerji yatırımlarının önceliklendirilmesinde yol gösterici olabilir. Gelecek çalışmaların, farklı bölgesel senaryolar ve uzun dönemli etkiler üzerinden bu yaklaşımların uygulanabilirliğini ve ölçeklenebilirliğini incelemesi önerilmektedir.
Context—Snake-like robots are biomimetic systems that can move effectively in narrow, complex, and restricted environments thanks to their modular and flexible body structures composed of numerous serially connected joints. These characteristics offer significant advantages, particularly in areas such as pipeline inspection, search and rescue operations, industrial maintenance applications, and exploration missions. The multiple degrees of freedom distributed along the body enable the robot to achieve high maneuverability but also make the control problem quite complex. Due to the dynamic interactions between segments, friction-based motion characteristics, and nonlinear system behavior, achieving reliable and accurate trajectory tracking emerges as a significant engineering problem.Objective—In this study, a reinforcement learning (RL) based control method has been developed to solve the trajectory tracking problem for snake-like robots in a two-dimensional plane.Method—In the proposed approach, the robot’s dynamic model was created in the Webots simulation environment, an open-source simulation program, and all training and testing processes were carried out in this environment. During the learning process, policy- based RL algorithms from the Stable-Baselines library were used. In this context, Proximal Policy Optimization (PPO) and three different RL algorithms were used during the training process. To enable the robot to adapt to different orientation scenarios, seven different angles defined in the range of +45 to −45 and trajectories of varying lengths were used. Thus, the goal was for the agent to learn a generalizable control policy not only for a specific trajectory type but also for tracks with different slopes and orientations.Results—The results obtained show that the PPO algorithm produced a higher average reward compared to other methods and exhibited a more stable learning process. After training was completed, the developed method was tested both on trajectories used during the training phase and on previously unseen trajectories. For the 0 trajectory, maximum errors were recorded as 0.093 m and 0.040 m for the x and y axes, respectively. Furthermore, the system exhibited robust generalization capabilities on a +22.5 trajectory, not encountered during the training phase, yielding maximum errors of 0.099 m and 0.052 m.Conclusion—These findings demonstrate that the proposed RL-based control approach can effectively solve the two-dimensional trajectory tracking problem in snake robots. In future studies, the proposed method can be extended to the three-dimensional trajectory tracking problem, or it can be evaluated under more complex conditions, such as scenarios involving obstacles.
Arka Plan—Akciğer kanseri, dünya genelinde kansere bağlı ölümlerin önde gelen nedenlerinden biri olup, hastalığın prognozu büyük ölçüde tanı anındaki evreye bağlıdır. Erken evrede teşhis edilen vakalarda sağkalım oranları önemli ölçüde artarken, mevcut tanı yöntemleri akciğer adenokarsinomunun moleküler heterojenliğini yeterince yansıtamamaktadır. Yüksek kapasiteli dizileme teknolojilerindeki gelişmeler, genomik ve klinik verilerin entegrasyonunu mümkün kılarak kanser evreleme ve prognoz tahmini için yeni fırsatlar sunmaktadır. Bununla birlikte, çoklu veri kaynaklarını etkin biçimde birleştiren ve yüksek doğruluk ile yorumlanabilirlik sağlayan hesaplamalı yaklaşımlara olan ihtiyaç devam etmektedir.Amaç—Bu çalışmanın amacı, gen ekspresyon verileri ve klinik bilgileri kullanarak erken evre (Evre I–II) ve ileri evre (Evre III–IV) akciğer adenokarsinomu ayrımını gerçekleştiren bütünleşik bir hesaplamalı çerçeve geliştirmektir. Çalışma ile hem sınıflandırma performansının artırılması hem de hastalık progresyonuna ilişkin biyolojik olarak anlamlı çıkarımlar elde edilmesi hedeflenmektedir. Ayrıca, önerilen modelin klinik karar destek sistemlerine entegre edilerek erken teşhis ve tedavi planlamasına katkı sağlaması amaçlanmaktadır.Yöntem—Çalışmada, TCGA-LUAD veri setinden elde edilen 193 hastaya ait RNA-seq gen ekspresyon verileri ve klinik bilgiler kullanılmıştır. Önerilen yaklaşım; ANOVA tabanlı özellik seçimi ile ayırt edici genlerin belirlenmesini, protein-protein etkileşim (PPI) ağı analizi ile biyolojik yapıların incelenmesini ve Cox orantılı Hazard modeli ile sağkalım analizinin gerçekleştirilmesini içermektedir. Ardından lojistik regresyon, destek vektör makineleri (SVM), rastgele orman ve gradyan artırma algoritmaları kullanılarak sınıflandırma işlemi yapılmıştır. Model performansı, tabakalı 5 katlı çapraz doğrulama yöntemi ile doğruluk, AUC-ROC, duyarlılık ve özgüllük metrikleri kullanılarak değerlendirilmiştir.Bulgular—Lojistik regresyon modeli %94.9 doğruluk ve %98.6 AUC-ROC değeri ile en yüksek performansı göstermiş, SVM modeli ise benzer sonuçlar elde etmiştir. Sağkalım analizi, ileri evre hastalığın anlamlı derecede daha yüksek mortalite riski ile ilişkili olduğunu ortaya koymuştur (HR = 6.50; p < 0.001). PPI ağı analizi sonucunda TP53, EGFR, KRAS, AKT1 ve PIK3CA genlerinin ağda merkezi rol oynadığı belirlenmiştir. Elde edilen sonuçlar, genomik ve klinik verilerin entegrasyonunun tek modaliteli yaklaşımlara kıyasla daha yüksek sınıflandırma başarısı sağladığını göstermektedir.Sonuç—Bu çalışma, akciğer adenokarsinomu evreleme ve prognoz tahmini için etkili ve yorumlanabilir birçok modlu yaklaşım sunmaktadır. Gen ekspresyon verileri ile klinik değişkenlerin birlikte kullanılması hem tahmin doğruluğunu hem de biyolojik anlamlılığı artırmaktadır. Önerilen çerçeve, klinik karar destek sistemlerine katkı sağlayabilecek potansiyele sahiptir. Gelecek çalışmalarda, bağımsız veri setleri ile doğrulama yapılması ve farklı veri türlerinin entegrasyonu önerilmektedir.
Context—In recent years, the increasing need for communication due to technological advancements, along with the rise in the number of manufacturers and users, has made the transition to 5th generation (5G) communication systems necessary. This transition to 5G communication has placed new requirements on antenna designs, necessitating the development of antennas with high bandwidth, a low profile, and the ability to support multiple users simultaneously. This paper focused on increasing the bandwidth of multiple input multiple output (MIMO) microstrip antennas and improving the isolation between ports.Objective—This article describes the design and fabrication of a low-profile, high-bandwidth, dual-polarization, four-port, 1×2 array microstrip MIMO antenna for systems operating at 5G frequencies. The designed antenna is manufactured, and its measurement results are compared with simulation results. To enable the proposed design to operate in dual polarization, two ports were placed perpendicular to the edges of the patch antenna. The patches are designed in a square shape to ensure the antenna has linear polarization and exhibits equal characteristics in both horizontal and vertical polarizations. It is aimed at the MIMO antenna to operate at a minimum of 25 GHz - 26 GHz range, linearly in dual polarization, and to have better than -20 dB isolation between ports despite the placement of patch structures with a distance 0.25λ. To increase the antenna bandwidth and improve isolation between ports, the defected ground structure (DGS) method and neutralization line method are employed in the design. The antenna is 2.4 cm × 2.96 cm in size and has 4 ports, achieves a bandwidth of 2.33 GHz with 7 dBi gain at two ports for horizontal polarization, and a bandwidth of 3.5 GHz with 9.18 dBi gain at two ports for vertical polarization. Additionally, the targeted isolation values between ports in the 25 GHz - 26 GHz range are achieved to be better than -20 dB. The designed MIMO antenna is also evaluated in terms of important MIMO parameters such as diversity gain (DG), envelope correlation coefficient (ECC), total active reflection coefficient (TARC), and channel capacity loss (CCL).Method—In the designed antenna, an RT/duroid 5880 substrate with a dielectric constant (εr) of 2.2, a loss tangent (tan δ) of 0.0009, and a thickness of 0.787 mm has been used. A 1 oz (0.035 mm) thick layer of copper was used on both the bottom (grounding plane) and top (radiation area) surfaces of the dielectric material.Results—To minimize the antenna's footprint, the patch structures were placed at a distance 0.25λ apart, while ensuring that the isolation between the ports remains below -20 dB. The bandwidths of the 4-port antenna were measured as 3.89 GHz, 4.72 GHz, 4.35 GHz, and 3.8 GHz respectively. Consistency was observed between the data obtained from measurements and simulation results.Conclusion—CST Microwave Studio simulation software was used for design and simulations.
Context—Paddy fields are prone to multi-matrix contamination because prolonged flooding can enhance exchanges between soil, water, and plant compartments, allowing pesticide residues to persist and potentially reach edible tissues. In Türkiye, evidence remains limited for stage-resolved monitoring that simultaneously links residue occurrence in environmental compartments and rice tissues with exposure and health-risk metrics in major rice-producing regions.Objective—This study investigated the occurrence, temporal variation, and health risk implications of pesticide residues in monoculture paddy fields in Ipsala and Biga (Türkiye) during the 2024 growing season.Method—Residues were extracted using the QuEChERS procedure and analyzed by LC–MS/MS across multiple matrices, including soil, water, shoot, stubble, husk, and harvested rice grain. Descriptive and association analyses were applied to characterize matrix-specific occurrence patterns and site-dependent co-occurrence among detected compounds. Matrix-to- matrix transfer was evaluated using concentration-ratio metrics to indicate empirical partitioning under flooded field conditions. Human health risks (carcinogenic and non-carcinogenic) for adults and children were assessed using the US EPA modeling framework.Results—Soil showed the highest residue occurrence, with detection frequencies of 45–52%, followed by notable detections in stubble and shoot. Temporal profiles differed by compound and site. In Biga, tridemorph remained elevated in soil and water and increased in husk and grain at harvest, whereas oxadiazon was largely associated with stubble at harvest with minimal levels in husk and grain. In Ipsala, clomazone showed a high pre-planting soil concentration followed by lower levels later in the season, and tridemorph decreased across harvest fractions (stubble: 40.7; husk: 29.5; grain: 21.7 µg kg⁻¹). Tridemorph and oxadiazon were detected in both regions despite being unauthorized in Türkiye, with tridemorph exceeding the maximum residue limit (MRL) in harvested rice grains. Correlation analysis suggested region-specific patterns, with a strong positive association between azoxystrobin and flutriafol in Biga (r = 0.97; p < 0.001) and a more complex co-variation structure was observed in Ipsala, which may reflect co-occurrence or matrix-specific distribution patterns. These active ingredients were also recorded in the pesticide application history of the study fields. Human health risk assessment results suggested that estimated non- carcinogenic and, where applicable carcinogenic risks for adults and children, calculated using the US EPA framework, remained within commonly applied regulatory acceptability thresholds.Conclusion—The findings suggest that pesticide residues may persist and partition differently across interconnected paddy matrices under field conditions and that unauthorized or legacy compounds can be detected, including in edible grain. Continued multi-matrix monitoring that considers stage-dependent behavior, together with source-to-receptor evaluation of transfer pathways, may support compliance checks and risk-informed management to improve the long-term sustainability of rice production in key Turkish rice-growing regions.
Context—Wind turbine towers are slender, flexible structures that are inherently susceptible to low-frequency vibrations induced by aerodynamic loading and rotor-related excitations. In particular, the overlap between structural natural frequencies and operational excitation ranges, such as 1P and 3P frequencies, may lead to resonance conditions that accelerate fatigue damage and reduce structural stability. Conventional vibration mitigation strategies, including tuned mass dampers (TMDs), are typically effective over a narrow frequency band and require precise tuning, which limits their robustness under varying operational conditions.Objective—In recent research, a metamaterial-inspired vibration mitigation approach based on the periodically distributed resonator idea has been investigated for the wind turbine towers with the specific emphasis on the mass ratio, tuning ratio and damping-related parameters.Method—The tower is modelled as a Euler-Bernoulli beam with a lumped nacelle mass, and the system is analyzed using a finite element formulation. Locally attached resonators are presented as mass-spring-damper systems and distributed along the tower height. A comprehensive parametric study is conducted to evaluate the influence of key design parameters, including the resonator tuning ratio, mass ratio, damping ratio, and the number of resonators. The dynamic response of the coupled system is assessed using frequency response function (FRF) within the operational frequency range of 0.1-1.5 Hz.Results—The results indicate that vibration attenuation is primarily governed by the frequency tuning of the resonators relative to the fundamental bending mode of the tower. The near-resonant configurations lead to increased interaction and partial suppression of the primary response peak. On the other hand, the off-tuned configurations contribute smoother response characteristics with limited direct influence on the dominant mode (around 3P). Increasing the resonator mass ratio enhances the interaction level; however, the overall attenuation remains constrained. Across all examined configurations, the observed peak reduction generally remains below 5%, indicating weak-to-moderate coupling between the resonators and the primary structure. The influence of damping is shown to introduce a trade-off between peak suppression and response stability, while increasing the number of resonators promotes more distributed interaction but does not significantly alter the magnitude of attenuation. The results further show that the system does not exhibit a distinct band gap, but rather a localized attenuation region.Conclusion—The proposed configuration is more appropriately interpreted as a distributed resonator system with metamaterial-inspired characteristics rather than a fully developed metamaterial structure. Overall, the findings provide a systematic assessment of a resonator-based vibration mitigation for wind turbine towers and highlight the limitations and potential of such systems for low-frequency vibration control in large-scale structures.
Arka Plan—Seyrek portföyler geniş bir yatırım varlığı uzayında oluşturulan, göreceli olarak az sayıda varlığın sıfırdan büyük ağırlığa sahip olduğu portföylerdir. Kardinalite kısıtlı ortalama-varyans modeli (KK-OVM) ve ortalama-mutlak-sapma modeli (KK-OMSM) matematiksel disiplinle oluşturulması için kullanılan matematiksel programlama modellerindendir. Bu modellerin en iyi çözümlerinden elde edilen portföylerin geçerliliği; modellerin varsayımlarının sağlanması, girdilerin doğası ve kalitesi, uygun istatistiksel tahmin metotlarının kullanımı ile ilintilidir.Amaç—Bu ampirik araştırmanın amacı kardinalite kısıtlı portföy optimizasyonu için kullanılan, ortalama varyans ve ortalama mutlak sapma risk ölçütleri ile yapılandırılmış iki matematiksel optimizasyon modelini, ürettikleri portföyleri farklı performans ve risk ölçütleri ile modeli varsayımlarının ihlal edilip edilmediğini kontrol eden, tahmin, çıkarım, hipotez testi adımlarını birleştiren bir yaklaşımla kıyaslamaktır.Yöntem—03/01/2021-04/01/2026 tarih aralığında toplanmış, BİST30 ve BİST100 endekslerinde listelenen hisse senetlerinin kapanış fiyatları kullanılarak ampirik bir çalışma yapılmıştır. Bu bağlamda; 1) Veri filtrelenmiştir. 2) Haftalık doğrusal ve logaritmik getiriler haftalık ortalama kapanış fiyatlarından hesaplanmıştır. 3) Logaritmik getirilerin normal dağılıma uyup uymadığının kontrolü Anderson-Darling hipotez testi ile sağlanmıştır. 4) Haftalık logaritmik getiri dağılımı geleceğe yansıtılmıştır. 5) Yansıtılmış logaritmik getiri dağılımından yansıtılmış doğrusal getiri dağılımı elde edilmiştir. 6) Örneklem Ortalaması Yakınlaştırılması ile doğrusal getiri senaryoları oluşturulmuştur. 7) Ortalama varyans ve ortalama mutlak sapma risk ölçütleri ile yapılandırılmış, kardinalite kısıtlı, iki matematiksel optimizasyon modeli yansıtılmış veri ile elde edilmiştir. Bu modeller farklı riskten kaçınma ve kardinalite parametreleri ile çözülmüştür. 8) Modellerin çözümü ile elde edilen portföylerden hesaplanan performans ölçütleri: yıllık getiri, yıllık Sharpe oranı, yıllık Sortino oranı, yıllık Calmar oranı, ortalama-mutlak-sapma oranı ve ortalama düşme oranı; risk ölçütleri: yıllık yarı-sapma, yıllık yarı-değişkenlik, yıllık sapma, yıllık değişkenlik, koşullu % 95 riskteki değer ve en büyük çekilmedir. Bütün portföyler için hesaplanan performans ve risk ölçütleri, istatistiksel olarak güven aralığı hipotez testleriyle kıyaslanmıştır.Bulgular—Birçok BİST100 hissesi için logaritmik getirilerin normalliğinin % 1 istatistiksel anlamlılıkta reddedilemediği tespit edilmiştir. Kardinalite kısıtlarındaki sıkılaşmanın etkin sınırı getiri-risk grafiğinde sağa kaydırdığı ve aynı getiri için daha fazla risk almak gerekliliği doğurduğu tespit edilmiştir.Sonuç—Kardinalite kısıtlı ortalama-varyans ile yapılandırılmış portföy optimizasyonu modellerinin, ortalama-mutlak-sapma ile yapılandırılmış benzer modellere göre daha düşük riskli portföyler ürettiği gözlemlenmiştir. Risksiz getiri oranının düşük olduğu durumda riske göre ayarlanmış getirilerinin daha yüksek olduğu fakat yüksek risksiz getiri oranları altında ortalama mutlak sapma modellerinin bazı durumlarda ortalama-varyans modellerinin önüne geçen portföyler ürettiği de gözlemlenmiştir.
Understanding the determinants of energy efficiency in electric public transport is critical for reducing operational costs and extending vehicle range under real-world traffic conditions. This study proposes a high resolution, traffic aware energy modeling framework for Electric Bus (E-Bus) systems using second by second (1 Hz) operational data collected from an urban route. By integrating traffic congestion metrics, vehicle dynamics, regenerative braking behavior, and machine learning based driving behavior analysis, the study provides a comprehensive assessment of energy consumption mechanisms beyond conventional average based evaluations. The results show that the average route level energy consumption of 3.3 kWh/km conceals substantial temporal variability primarily driven by congestion induced stop and go dynamics. Traffic congestion increases energy consumption by up to 22%, not merely by reducing speed but by amplifying acceleration variance and braking frequency. To maintain the optimal driving range, approximately 78.32% of total braking energy must be recuperated through regenerative braking, a requirement that is strongly conditioned by traffic state and driving behavior. Although driver to driver differences in total energy consumption appear moderate (6–9%), the underlying control strategies differ significantly, leading to cumulative long term impacts at fleet scale. Under identical traffic conditions, autonomous driving reduces total energy consumption by 11–14% and increases regenerative braking utilization by 9–12% compared to human driven operation by enforcing smoother longitudinal control. The findings demonstrate that energy efficiency in E-Bus systems is jointly governed by traffic dynamics and control behavior, highlighting the necessity of behavior aware driving strategies and autonomous control for next generation electric public transport systems.
This study aims to optimize the hybrid RFW–MLM process for AA6063 aluminum alloy using Response Surface Methodology (RSM). The effects of rotational speed, forging pressure, and MLM geometric parameters on tensile strength and microstructural evolution were statistically modeled and experimentally validated to determine the optimal processing window. Rotary friction welding (RFW) has been widely utilized across diverse industries, including aerospace and automotive, as a significant method for producing robust joints. This approach is particularly advantageous for materials such as aluminium alloys, which exhibit a heightened susceptibility to heat-induced complications during welding. Nevertheless, the investigation of other methods has been motivated by concerns regarding faults caused by heat, leading to increased importance of the mechanical locking method (MLM). This study investigates the intersection of combined RFW and MLM to elucidate the manner in which these processes interact to affect the microstructural attributes and mechanical behaviors of welded joints. The main aim of this study was to thoroughly assess the quality of joints made from Al6063 aluminium using two different approaches. By employing the response surface methodology, it was endeavored to optimize the process parameters of the friction welding apparatus to attain the highest tensile strength. To validate the model, Analysis of Variance (ANOVA) was employed, confirming its significance with an F-value of 34.98. The rotational velocity and forging force collectively contributed to 24% of the overall impact, emphasizing their critical role in joint strength. The normal distribution of residuals further substantiated the model's adequacy. Microstructural analysis revealed phase transformations, including the formation of S, T, PFZ (Power Frequency Zone), and Guinier–Preston–Bagaryatsky (GPB) phases, which influence the mechanical properties of welded joints. Notably, friction-induced overheating in the S12 group led to T-phase formation, potentially reducing strength. The PFZ phase observed in the S8 group affected the alloy’s microstructure, while GPB zones in the S12 group affected hardness and strength. These findings provide compelling evidence that the concurrent use of RFW and MLM techniques significantly enhances the mechanical characteristics of AA6063 aluminium, with microstructural transformations playing a crucial role in determining joint performance
Ekonomik kalkınma, teknolojik gelişmeler ve nüfus artışı, enerji gereksinimlerini daha da artırmaktadır. Bu nedenle, enerji tüketiminin doğru ve güvenilir yöntemlerle öngörülmesi ile uygun enerji üretim yöntemlerinin belirlenmesi, etkin ve sürdürülebilir enerji planlaması açısından büyük önem taşımaktadır. Bu çalışmada görece düşük güneş ışınımına sahip bölgede parabolik oluk güneş kolektörlerin (POGK) uygulanabilirliği incelenmiştir. Amaç— Bu çalışmada; Kastamonu Üniversitesine ait yerleşkenin 2040 yılı enerji talebinin doğru ve güvenilir yöntemlerle belirlenmesi, belirlenen talebin tasarlanan optimum kapasitedeki POGK enerji üretim santralinden karşılanması ve santral enerji verimliliğinin artırılması amaçlanmıştır. Yöntem—Bu çalışmada, yapay sinir ağı (YSA) modellerinin geliştirilmesi amacıyla MATLAB kullanılarak bir yazılım oluşturulmuştur. Çalışma kapsamında geliştirilen tüm Çok Katmanlı Algılayıcı (MLP) modelleri dört katmanlı bir mimariye sahiptir. Giriş katmanındaki nöron sayısı ile gizli katmandaki nöron sayısı 1 ile 10 arasında değiştirilerek toplam 100 farklı YSA modeli elde edilmiştir. Veri setine uygun Box–Jenkins modelinin belirlenmesi amacıyla hem elektrik hem de doğal gaz tüketim verileri için çeşitli modeller geliştirilmiştir. Bu modeller arasından, elektrik tüketim verileri için sabit terim içermeyen ARIMA (3,1,2) modeli; doğal gaz tüketim verileri için ise sabit terim içeren ve doğal logaritmik dönüşüm uygulanmış ARIMA (1,0,0) modeli uygun bulunmuştur. Yerleşkenin enerji talebi, elektrik ve termal enerji üretimini birlikte gerçekleştiren POGK santralinden enerji üretimi yöntemiyle karşılanmıştır. Bulgular—Elektrik ve doğal gaz tüketim verileri için çok katmanlı algılayıcı (MLP) modeli kullanılarak elde edilen Ortalama Mutlak Yüzde Hata (MAPE) değeri %4,33 olarak hesaplanırken, aynı veriler için kurulan ARIMA modelinde bu değer %4,94 olarak bulunmuştur. Her iki veri setinde de daha düşük hata oranlarına sahip olan YSA tabanlı modellerin, ARIMA modellerine kıyasla daha yüksek tahmin doğruluğu sunduğu belirlenmiştir. Elde edilen model sonuçlarına göre, 2040 yılı itibarıyla yerleşkenin elektrik enerjisi talebinin 3.393.349 kWh, doğal gaz talebinin ise 1.094.804 Nm³ düzeyinde olacağı öngörülmektedir. Bu talebin karşılanabilmesi amacıyla, her biri 24 adet POGK’lerinden oluşan toplam 15 modülden meydana gelen bir POGK santrali tasarlanmıştır. Santral, yerleşkenin toplam elektrik enerjisi ihtiyacının tamamını ve ısıl enerji gereksiniminin yaklaşık %52’sini karşılayabilecek kapasitede olduğu belirlenmiştir. Sonuç—POGK sistemi, yenilenebilir enerji kaynaklarının etkin kullanımını destekleyerek enerji arz güvenliğinin artırılmasına yönelik önemli bir çözüm sunmaktadır. Araştırma bulguları, güneşlenme süresinin görece düşük olduğu bölgelerde dahi bu tür sistemlerin hem elektrik hem de termal enerji üretiminde uygulanabilir olduğunu hesaplamalar ve analizler aracılığıyla ortaya koymaktadır.
Context—Comparative analysis of health indicators at national and international levels is essential for identifying regional disparities and evaluating differences in healthcare access, quality, capacity, and outcomes. Clustering analysis is widely used to group regions with similar health profiles and to support evidence-based regional health planning. In Türkiye, regional inequalities in health indicators persist, yet related studies remain fragmented in scope and methodology. Despite increasing data availability and analytical capacity, no comprehensive review has examined regional clustering studies conducted in Türkiye. A structured synthesis is therefore required to clarify methodological trends, dominant practices, and unresolved gaps in this field, and to assess how existing research contributes to regional health policy development.Objective—The objective of this study is to present the first comprehensive review of regional clustering studies in Türkiye based on health indicators. The review aims to inform future research directions and assist policymakers in developing data-driven and regionally targeted healthcare strategies.Method—A comprehensive review was conducted to identify studies on the clustering of provinces and regions in Türkiye based on health indicators. Literature published up to 2025 was retrieved from Google Scholar using relevant keywords in both English and Turkish. The study examines publication trends, geographical coverage, clustering techniques, methods used to determine the number of clusters, selected health indicators, data sources, study periods, and software tools. Results—Although the number of studies has increased notably in recent years, the overall research volume remains limited given the importance of the subject. Moreover, the predominant reliance on conventional clustering methods highlights the need for greater methodological diversity, interdisciplinary collaboration, and the integration of advanced algorithms to strengthen the validity and practical relevance of regional health analyses.Conclusion—This review highlights methodological concentration in existing studies. Expanding methodological diversity is essential to enhance analytical robustness and policy relevance. These improvements will support more accurate health assessments and contribute to equitable healthcare planning in Türkiye.
In the contemporary business environment, characterized by intense competition and rapid global change, effective and sustainable supply chain management is of paramount importance. Organizations rely on robust closed loop supply chains (CLSCs) to efficiently manage material and cash flows while integrating recycling and reuse processes to promote sustainability. The present study proposes a novel optimization model for order allocation in a CLSC to facilitate the design of a sustainable supply chain network. The proposed scenario based mixed integer linear programming model addresses supplier capacity uncertainty based on the disruptions exacerbated by the COVID 19 pandemic. On the one hand, the network design presented focuses on sustainable closed-loop supply chains for perishable products, proposing a model that monitors the time products spend in the warehouse and then manages their dispatch to the recycling center.The text also highlights the strategic advantage of evaluation of many suppliers to increase flexibility and responsiveness in times of supply chain disruption. The model incorporates a supplier score parameter that integrates criteria based on Lean, Agility, Resilience, and Green (LARG) principles and pandemic related challenges through evaluations of existing suppliers. A comprehensive literature review investigates the factors influencing supplier selection processes. The experts assess the importance of the criteria presented in the supplier selection by applying a series of judgments together with the selection criteria that form the basic framework. Each judgment is objectively evaluated using the Full Consistency Method (FUCOM). Furthermore, the experts are evaluated according to their experience and the Intuitionistic Fuzzy Sets (IFS) method determines the influence of their opinion on the decision. This comprehensive approach aims to optimize network performance while supporting long term sustainability and responsiveness in a post pandemic world. The proposed model is tailored to the healthcare sector and applied to the network design of a hospital in Türkiye. The Complex Proportional Assessment Method (COPRAS) was employed to evaluate and score suppliers, with these scores integrated into the proposed mathematical model. Sensitivity analysis is conducted on key factors, including supplier scores, product shelf life, and capacity. The results highlight the pivotal role of supplier selection in optimizing closed loop supply chain design. The study emphasizes optimal inventory tracking to manage perishable goods, underscoring its critical role in achieving sustainable and efficient healthcare operations.
This study investigates the determinants of electric vehicle (EV) pricing through an interpretable machine learning framework. Using XGBoost on a comprehensive dataset of technical specifications, we achieve high predictive accuracy (R²=0.958) while employing SHAP and LIME to deconstruct the model's decision logic. Our analysis reveals a fundamental shift in automotive valuation principles: integrated performance characteristics synthesized via Principal Component Analysis. It emerges as the primary price driver, followed by vehicle dimensions, which exhibit non-linear threshold effects. Notably, traditional differentiators had minimal impact, suggesting that electric powertrains are redefining conventional automotive hierarchies. The complementary interpretability methods consistently demonstrate that EV pricing rewards engineering substance over traditional status markers, with performance bundles and dimensional thresholds creating clear market stratification. These findings provide manufacturers with quantifiable engineering targets for premium positioning and offer consumers unprecedented transparency into feature valuation. The study establishes that interpretable machine learning not only predicts prices but also uncovers the emerging economic logic governing the electric vehicle revolution, where technological integration and physical proportions supersede historical automotive status symbols.
Vehicle Edge Computing (VEC) is a computing paradigm specifically designed to support the execution of computationally intensive vehicle applications while ensuring low latency, efficient bandwidth utilization, and reduced energy consumption. In modern vehicular environments, vehicles are required to process large amounts of data generated by diverse applications, and local onboard resources are often insufficient to meet these demands. Therefore, vehicles must efficiently offload computational tasks to external computing systems that offer higher processing capabilities in order to maintain optimal computational performance, especially under dynamic and changing network conditions.This paper investigates a VEC scenario in which vehicles outsource their computational tasks with the objective of optimizing overall computation time and increasing the successful task transfer rate. In such scenarios, several factors affect computational efficiency, including heterogeneous task requirements and the inherent mobility of vehicles. Variations in task size, delay sensitivity, and processing demands, combined with frequent changes in vehicle location, can significantly affect the performance of task offloading mechanisms and resource utilization.To address these challenges, the study incorporates and utilizes a fifth-generation (5G) radio network to effectively manage task offloading decisions and computational resources. A 5G-based task offloading scheme is proposed, aiming to manage computational resources in a mobility-aware manner. By considering vehicle mobility, the proposed scheme adapts task export decisions to changing network conditions, thereby improving system robustness and efficiency. The scheme is further supported by a distributed communication model, which enables near-optimal solutions by coordinating task offloading and resource allocation across the network.Furthermore, the proposed approach integrates a fifth-generation new radio (NR) communication model to enhance system performance. This model includes both cellular connectivity and millimeter wave (mmWave) communication, allowing the system to benefit from the strengths of each communication mode. Simulation results demonstrate that the proposed model significantly improves computational efficiency, particularly in terms of successful task handover and Quality of Experience (QoE). QoE represents the overall service performance as perceived by users and is influenced by key factors such as latency, throughput, and error rates. Overall, the proposed approach achieves improved task offloading efficiency, reduces task failure rates, and enhances QoE through the effective integration of mmWave and 5G NR communication models.
Arka Plan—Sera gazı ve özellikle CO2 emisyonlarındaki artışın yarattığı çevresel sorunlara ilişkin farkındalığın artması sonucunda, CO2 Yakalama ve Depolama teknolojilerinin uygulanmasıyla birlikte fosil yakıtlara alternatif yakıtların kullanımına dayanan ve sürdürülebilirliği temel alan verimli sistemlerin geliştirilmesi ihtiyacı ön plana çıkmıştır. Karbon köpükler, hafiflikleri ve eşsiz özellikleri nedeniyle birçok uygulamada geleneksel malzemelere karşı önemli bir alternatif olarak kullanılan ileri teknoloji malzemeler arasında yer almaktadır. Biyokütle-esaslı karbon köpüklerin CO2 adsorpsiyonunda değerlendirilmesi hem ucuz ve bol bulunan kaynakların kullanımını sağlaması hem de CO2 yakalanma süreçlerinin geliştirilmesine katkı sunması açısından üzerinde çalışılması gereken önemli bir araştırma alanıdır.Amaç—Bu çalışmanın amacı, geleneksel fosil yakıt türevi karbon köpüklere kıyasla daha çevre dostu bir yol sunan, solvolitik sıvılaştırma reaksiyonu ile bu malzemelerin sentezinde pirolitik yağın alternatif bir çözücü olarak kullanılabilirliğini inceleyerek biyokütle-esaslı karbon köpüklerin CO2 adsorpsiyon davranışının izoterm ve istatistiksel analiz ile değerlendirilmesidir.Yöntem—Bu çalışmada, gürgen talaşının solvolitik sıvılaştırılmasıyla üretilen biyopoliol-esaslı karbon köpüklerin CO2 adsorpsiyon performansı izoterm modelleme ve istatistiksel analiz yaklaşımlarıyla değerlendirilmiştir. Bu kapsamda, çözücü türü, çözücü:biyokütle oranı ve kimyasal aktivasyon olmak üzere sentez parametrelerinin CO2 adsorpsiyon kapasitesi üzerindeki etkileri 23 tam faktöriyel deney tasarımına dayalı istatistiksel analiz ile incelenmiştir. Ayrıca, denge adsorpsiyon davranışının ve adsorpsiyon mekanizmasının aydınlatılması için Langmuir, Freundlich, Dubinin-Radushkevich ve Temkin modellerine göre izoterm analizinin yapılması ile sentez koşulları ve adsorpsiyon performansı arasındaki ilişki ortaya konulmuştur. Adsorban olarak kullanılan karbon köpüğün yüzey heterojenliğinin, farklı enerji seviyelerine sahip adsorpsiyon bölgelerinin adsorban-adsorbat afinitesi üzerindeki etkileri aydınlatılmıştır.Bulgular—İzoterm modelleme sonuçları, özellikle kimyasal aktivasyon uygulanmış karbon köpüklerde Freundlich modelinin daha yüksek uyum katsayıları (R^2 > 0,99) sergileyerek deneysel verileri en iyi şekilde yansıttığını göstermiştir. İstatistiksel analiz sonucu oluşturulan regresyon modeli, kimyasal aktivasyonun CO2 adsorpsiyon kapasitesini artıran en baskın faktör olduğunu, çözücü türü ve çözücü:biyokütle oranının ise orta derecede doğrudan etkileri olduğunu ortaya koymuştur. İkili etkileşim terimleri, karbon köpük üretim parametreleri arasında sinerjik farklar olduğunu, çözücü türü ile ilgili kombinasyonların pozitif etki sağladığını, buna karşın yüksek çözücü:biyokütle oranının aktivasyon verimliliğini potansiyel olarak azalttığını göstermiştir. Bu bulgular, gözenek gelişimi ve adsorpsiyon performansı arasındaki etkileşimi vurgulayan yapısal karakterizasyonla uyumluluk sergilemektedir.Sonuç—Bu çalışma, karbon köpük sentez parametrelerinin CO2 adsorpsiyon kapasitesi üzerindeki etkisini adsorpsiyon mekanizması ve istatistiksel anlamlılık düzeyinde açıklayarak karbon köpüklerin uygulama odaklı kullanım potansiyellerine yönelik özgün bir yaklaşım sunmaktadır. Karakterizasyon ve istatistiksel analizlerin ilişkilendirilmesine göre, sınırlı deney sayısına sahip faktöriyel tasarımda %90 güven seviyesinin kullanılarak potansiyel etkilerin ön eleme amacıyla değerlendirilmesinin uygun olduğu ve kimyasal aktivasyonun optimizasyon çalışmalarında öncelikli olarak ele alınması gerektiği sonucuna ulaşılmıştır.
Nowadays, numerous studies focus on Wi-Fi-based indoor positioning, primarily utilizing RSSI (Received Signal Strength Indicator) and FTM (Fine Time Measurement). These studies aim to mitigate noisy results caused by physical obstacles and hardware limitations. Beyond standard noise, edge cases like connection loss represent critical environmental dynamics. The rarity of such disconnections and packet loss events complicates analytical investigation and results in limited literature. However, research suggests that these anomalies are driven by environmental factors rather than being purely random events.This study provides a statistical and machine learning-based evaluation to examine the relationship between signal behaviors and packet loss anomalies across different locations. A dataset from an industrial automation laboratory is utilized to analyze packet losses within RSSI and FTM-based distance measurements. A classification and Moran’s I-based framework is developed to distinguish anomaly states from normal operating conditions.To analyze regional packet loss behaviors, time series data are collected with equal sampling at each location. Features such as RSSI Slope and FTM Distance Error Slope are extracted using a sliding window approach. Relationships are analyzed via Pearson and Spearman correlation coefficients, with significance validated through permutation tests. Anomaly detection is framed as a supervised learning problem, utilizing RandomizedSearchCV and Random Under Sampler. Performance is evaluated using Logistic Regression, Random Forest, XGBoost, and Isolation Forest models. Additionally, Local Moran’s I is applied to analyze regional spatial relationships.Permutation tests revealed that while Pearson correlations are statistically significant, the magnitude of the relationship remained limited. Under normal conditions, the standard deviation of correlation values for RSSI and FTM features ranged between 0.0708–0.0855. In contrast, for packet loss anomalies, this standard deviation increased approximately 4–6 fold, reaching levels between 0.3898 and 0.4163. In supervised learning trials, XGBoost achieved the highest performance with 0.7351 Accuracy and a 0.7348 Macro-F1 score, whereas Isolation Forest showed the lowest effectiveness. Regional analysis yielded a Moran’s I score of 0.20 for FTM distance error-based features.The findings demonstrate that packet loss events are not random but exhibit systematic behaviors linked to electromagnetic and structural environmental properties. The marked increase in correlation values during anomalies shows that packet loss can be characterized both temporally and regionally. Results indicate that tree-based methods capture these anomalies more effectively than linear models. Furthermore, Local Moran’s I analysis confirms that packet loss behaviors possess regional dependency and spatial clustering.
Context— Time series classification (TSC) plays a critical role in many application domains such as healthcare, fault diagnosis, and signal analysis, where signals are often non-stationary and noisy. Although convolutional neural networks (CNNs) have achieved strong performance, their effectiveness is limited when operating directly on raw signals, particularly in capturing multi-frequency characteristics. Image-based representations such as GAF, MTF, and RP have enabled CNNs to exploit temporal structures more effectively. However, most existing approaches generate these images directly from raw signals, leaving the potential of frequency-aware representations underexplored.Objective— The objective of this study is to develop a hybrid time series classification framework tailored for sensor-based signals with rich frequency content. In particular, the study investigates whether generating GAF, MTF, and RP images from different Discrete Wavelet Transform (DWT) components can enhance the representation quality of frequency-dependent time series. By assigning distinct image encodings to approximation and detail coefficients, the proposed approach aims to improve dataset representations by capturing complementary information across multiple frequency bands commonly observed in sensor data. The effectiveness of the framework is evaluated on benchmark datasets from the UCR archive.Method— In the proposed method, each time series is decomposed using a two-level Discrete Wavelet Transform with the db4 wavelet to obtain approximation and detail coefficients. These coefficients are then independently transformed into image representations: MTF from approximation coefficients, GAF from medium-frequency components, and RP from high-frequency components. The resulting images are combined into multi-channel inputs and classified using a lightweight CNN architecture. Additional experiments using direct image transformation and multi-branch and multi-channel CNN designs are conducted to assess the contribution of wavelet-based representations. Results— Experimental results on multiple UCR benchmark datasets demonstrate that the proposed DWT-based image representations improve classification accuracy, particularly for long and noisy time series. The wavelet-based approach shows clear advantages in datasets characterized by complex temporal dynamics and low signal-to-noise ratios. For short and relatively clean signals, CNNs trained on raw image representations achieve comparable or slightly better performance. These findings indicate that frequency-aware image encoding enhances robustness without increasing model complexity.Conclusion— This study shows that integrating Discrete Wavelet Transform with image-based time series representations provides a more balanced and informative feature space for CNN-based classification. By mapping different DWT components to GAF, MTF, and RP, the proposed framework captures temporal patterns across multiple frequency scales. The results highlight the importance of frequency-specific encoding, especially for challenging real-world signals. Future work may explore deeper architectures and adaptive wavelet-image assignments to further improve classification performance.
Context— Fused Deposition Modeling (FDM) has become a commonly adopted additive manufacturing method because it offers economical production, geometric versatility, and broad material availability. Among the thermoplastics used in FDM, polylactic acid (PLA) is widely preferred due to its environmentally friendly nature and favorable processing characteristics. Nevertheless, the mechanical properties and surface finish of PLA parts produced by FDM are highly sensitive to both manufacturing parameters and post-processing practices. Although previous research has reported the separate effects of factors such as layer thickness, infill density, and annealing, studies that systematically evaluate their combined impact are still scarce.Objective— This research focused on examining how variations in layer thickness, infill density, and annealing temperature affect the tensile performance and surface characteristics of PLA parts produced by FDM. In particular, the study aimed to determine which processing parameters most strongly influence tensile strength, elastic modulus, elongation, specific tensile strength, and surface roughness, as well as to explain their role in balancing stiffness and ductility in printed PLA parts.Method— An L9 Taguchi orthogonal array was implemented to assess the influence of layer thickness set at 0.12, 0.16, and 0.20 mm, infill density levels of 20, 40, and 60 percent, and annealing states consisting of as-printed, 60 °C, and 90 °C. Tensile samples made of PLA were fabricated using an FDM-based 3D printing system under controlled processing conditions. Mechanical characterization was carried out through tensile testing to obtain tensile strength, elastic modulus, and elongation values, whereas specific tensile strength was determined by incorporating sample mass measurements. Surface quality was evaluated by measuring the average surface roughness (Ra) with a three-dimensional optical profilometer. The effects and relative significance of the selected parameters were statistically analyzed using signal-to-noise ratios and analysis of variance techniques.Results— The results showed that infill density was the most important factor affecting tensile strength, with a contribution ratio of 87.76%, and maximum strengths of approximately 43–45 MPa obtained at 60% infill. Layer thickness was identified as the dominant parameter controlling elastic modulus (48.17%) and elongation (71.64%), its critical role in the stiffness–ductility balance. Surface roughness increased as thicker layers formed more pronounced and visible layer-step structures. Surface roughness increased as thicker layers produced more pronounced and visible layer-step structures. The overall effect of layer thickness on surface roughness was 84.73%.Conclusion— This work presents an integrated assessment of how multiple processing parameters influence the mechanical behavior and surface characteristics of PLA components. The results support the optimization of strength, ductility, weight efficiency, and surface finish, and contribute practical insight for the engineering design of PLA parts manufactured by FDM.
Context— Phenolic compounds from plant sources are widely recognized for their antioxidant and antidiabetic properties and are increasingly investigated for functional food and nutraceutical applications due to their potential role in the prevention of oxidative stress–related metabolic disorders. In recent years, increasing attention has been directed toward plant leaves as alternative and sustainable sources of bioactive compounds. Leaves of Sorbus domestica L. (service tree) represent an underutilized plant material with considerable potential bioactive value; however, information on efficient extraction strategies and functional characterization of their phenolic compounds remains limited. In particular, comparative evaluations focusing on both extraction efficiency and biological activity are still insufficient. Therefore, systematic comparisons of traditional extraction and ultrasound-assisted extraction methods are required to better understand their effectiveness in recovering phenolic compounds from S. domestica leaves.Objective— The aim of this study was to compare traditional extraction (TDE) and ultrasound-assisted extraction (UAE) methods for the extraction of phenolic compounds from S. domestica leaves. For each method, extraction conditions were optimized, and the resulting extracts were evaluated in terms of phenolic content (TPC and TFL), antioxidant activity, and antidiabetic potential.Method— S. domestica L. leaves were used as the plant material in this study. Phenolic compounds were extracted using traditional extraction and UAE methods, with extraction temperature, ethanol concentration, and extraction time selected as independent variables. An experimental design approach was employed to optimize the extraction conditions for both techniques. Total phenolic content (TPC) and total flavonoid content (TFL) were quantified using spectrophotometric methods. Antioxidant activities of the extracts were evaluated using ABTS, FRAP, and DPPH assays, which are commonly applied to assess different antioxidant mechanisms. In addition, the antidiabetic potential of the extracts was assessed through α-amylase and α-glucosidase inhibition tests.Results— The optimal conditions for TDE were 49.63% ethanol, 70.0 ⁰C, and 66.32 min, whereas UAE achieved optimal performance at 53.43% ethanol, 69.04 ⁰C, and 85.38 min. Under optimized conditions, UAE produced 94.38 mg GAE g-1 TPC and 89.13 mg QE g-1 TFL, whereas TDE yielded 73.21 mg GAE g-1 and 68.77 mg QE g-1, respectively. This corresponds to approximately 29% higher TPC and 30% higher TFL in UAE compared to TDE, clearly demonstrating the superior extraction efficiency of the ultrasonic method. Antioxidant activities were also enhanced under UAE conditions. For example, UAE improved α-amylase inhibition by approximately 33% (IC50 reduced from 50.29 to 33.86 mg mL-1) and α-glucosidase inhibition by approximately 12% compared to TDE, indicating stronger enzyme inhibitory potential. The lower IC50 values observed under UAE conditions suggest improved biological functionality of the extracted phenolics, which is particularly relevant for modulating postprandial hyperglycemia.Conclusion— S. domestica leaves are a rich source of phenolic compounds with considerable antioxidant and antidiabetic potential. The quantitative improvements achieved by UAE indicate that extraction strategy significantly influences both yield and bioactivity. These findings highlight the potential application of UAE-derived S. domestica leaf extracts as functional ingredients in nutraceutical formulations, functional beverages, and glucose-regulating food systems.
