In this study, a reinforcement learning (RL)–based route planning strategy is proposed to enable uncrewed aerial vehicles (UAVs) to reach a target point within a grid-based mission area with optimal time and energy efficiency. Using the Proximal Policy Optimization (PPO) algorithm, alongside a Deep Q-Network (DQN) baseline for comparative analysis, the UAV agent learns to minimize mission duration and corresponding battery energy consumption. The model was trained in a 10×10 grid environment, utilizing an energy-aware reward shaping structure. During training, the PPO agent demonstrated strong performance, achieving near-perfect success rates and reaching the target in efficient steps, while outperforming the DQN baseline in terms of stability. Overall, the results indicate that the proposed PPO model effectively learns environmental dynamics without relying on inefficient “wait” actions, leading to a robust and energy-efficient navigation policy.
Accurate forecasting of CO₂ emissions is essential for climate policy development, sustainable energy planning, and long-term environmental management. However, many existing studies are limited by single-country analyses, short temporal coverage, or reliance on a single forecasting methodology. To address these limitations, this study presents a comparative forecasting framework based on statistical and neural-network-based approaches for estimating annual CO₂ emission trends in six major emitting economies: China, the United States (USA), the European Union (EU27), India, Russia, and Japan. Using annual CO₂ emissions data obtained from the EDGAR database for the period 1970–2021, several statistical models (AUTO.ARIMA, HOLT, and TBATS) and neural-network-based approaches (NNETAR, MLP, and ELM) were comparatively evaluated. Forecasting performance was assessed using Mean Absolute Percentage Error (MAPE) and Mean Absolute Scaled Error (MASE). The results demonstrate that model performance varies significantly depending on the emission dynamics of each country. Neural-network-based approaches exhibited strong predictive capability for rapidly changing and nonlinear emission trends, particularly in China and Russia, whereas statistical models such as TBATS and AUTO.ARIMA provided competitive forecasting accuracy for more stable emission patterns observed in the USA and Japan. Among all evaluated approaches, TBATS achieved the highest forecasting performance for India and the USA, while MLP and NNETAR models produced robust results for China and EU27. This study contributes to the literature by providing a long-term cross-country comparative evaluation of statistical and neural-network-based forecasting approaches for CO₂ emissions. The findings offer an evidence-based analytical framework that can support policymakers in energy transition planning, emission reduction strategies, and sustainable development policies aligned with global climate objectives.
Türkiye, due to its geostrategic location, is exposed to various disaster risks. Bursa, which lies on the North Anatolian Fault Line, is particularly vulnerable to earthquakes as well as floods, landslides, and erosion. Therefore, preparedness activities carried out before disasters are as crucial as post-disaster interventions. In disaster logistics, the main objective is to ensure that the needs are met quickly and reliably through the proper location of logistics centers. In this study, the Interval Type-2 Fuzzy TOPSIS method was applied to determine the most suitable disaster logistics center location for Bursa. Based on literature review and expert opinions, criteria and alternative centers were identified and ranked using the method. Sensitivity analysis was then conducted to test the reliability of the model, and the results indicated that Yunuseli is the most appropriate location. The findings contribute significantly to minimizing potential loss of life and property, ensuring rapid response to basic needs, and optimizing logistics costs during disasters.
In this study, test specimens were produced using the Fused Deposition Modeling (FDM) method with Polylactic Acid (PLA) material at a constant layer height (0.1 mm) and scanning strategy (Concentric), with varying extrusion temperatures (205, 215, 225, and 235 °C) and different printing speeds (50 mm/s and 100 mm/s). Within the scope of the experimental design, the effects of production parameters on mechanical and physical properties (Shore D hardness, density, and tensile strength) were investigated. According to the analysis results, increasing the extrusion temperature to 235 °C enhanced the melt flowability of the material, thereby increasing interlayer bond strength and minimizing the structural void ratio. When evaluating kinematic parameters, it was determined that specimens produced at a printing speed of 50 mm/s exhibited a more homogeneous matrix structure compared to 100 mm/s printing speed and increased tensile strength by approximately 12%. Conversely, at high printing speeds such as 100 mm/s, insufficient interlayer thermal welding time weakened the structural integrity, causing a significant decrease in hardness values. The statistical reliability of the obtained data was verified by the Anderson-Darling normality test, with the P-Value found to be above the 0.05 significance level and the AD statistic below the critical threshold of 0.752. In conclusion, it was determined that proper optimization of thermal and kinematic parameters plays a critical role in maximizing the mechanical performance of components produced by FDM.
This study presents a high-fidelity numerical investigation into the enhancement of turbulent heat transfer and flow characteristics within a rectangular channel equipped with novel trapezoidal diversion plates. The research systematically evaluates the impact of baffle count and geometry on thermal-hydraulic performance at a high Reynolds number of 87,300. A critical contribution of this work is the comprehensive performance assessment of five turbulence models: Standard k-ε, Standard k-ω, SST k-ω, Spalart–Allmaras, and Reynolds Stress Model (RSM). Validation against experimental data reveals that the RSM and SST k-ω models demonstrate superior predictive accuracy in capturing complex flow physics, with a maximum velocity deviation of only 3.8%. Quantitative findings indicate that increasing the number of trapezoidal baffles induces a jet-like flow acceleration, reaching peak velocities of 13.8 m/s (a 77% increase over inlet velocity) and intensifying turbulence-driven mixing. This structural modification results in a significant increase in the Nusselt number, ranging from 14% to 22% compared to standard configurations. The results fill a prominent gap in literature by providing precise numerical data on the synergy between trapezoidal geometries and second-moment closure models, offering a robust framework for the design of next-generation industrial heat exchangers.
This study investigates the static bending behavior of cantilever sandwich beams with a porous core structure and fiber-reinforced composite outer layers. Under a vertically concentrated load applied to the middle span of the beam, the differential equations governing the system were solved in a fully analytical and closed-form manner using the Laplace transform method. Within the scope of the research, the coupling effects of core porosity coefficient, core-surface layer thickness ratio, outer layer fiber orientation angle, material orthotropy degree on displacement were systematically mapped. Numerical findings showed that after the slenderness ratio exceeds a certain threshold, the slenderness effect aggressively increases the deflection values, and porosity control becomes critical in thick-core designs. Furthermore, it is observed that the structure's resistance to section thickening weakens as the fiber angle moves away from the longitudinal axis. The precise results obtained constitute a unique and flawless design guide for the optimization processes of lightweight sandwich structures in the aerospace and defense industries.
The aim of this study is to design, model, and thermo-economically evaluate a PV-battery-grid integrated reverse osmosis desalination system for geothermal brine treatment in the Balcova Geothermal District Heating System, İzmir, Türkiye, with the objective of maximizing photovoltaic energy utilization and minimizing grid dependence. The system comprises PV panels, batteries, and the grid to supply electricity during the desalination process and it is modelled and simulated via TRNSYS software. A comprehensive thermo-economic modelling approach is employed to evaluate system performance under three solar-radiation scenarios, namely the low-radiation (LR), average-radiation (AR), and high-radiation (HR) scenarios. The findings indicate that the scenario with minimal radiation has the maximum net present value and the longest payback period. The scenario with the average radiation is identified as the most feasible investment, with a net present value of $52,218.7 and a payback period of 11.1 years. The cost of energy varies depending on the scenario, with the lowest radiation resulting in the lowest cost per kWh at $0.22/kWh. However, taking into account the payback period of 11.16 years and a net present value of $52,218.70, the average radiation scenario is the most viable investment. Overall, the findings reveal that the proposed system offers a sustainable and economical solution to water scarcity, with detailed energy, exergy, and economic analyses supplying vital information regarding system design and scalability.
This study presents the design and performance assessment of a grid-connected solar photovoltaic (PV) system installed at the National Institute of Textile Engineering and Research (NITER), Bangladesh. Using PVsyst simulations with local climatic data, the technical feasibility, economic viability, and environmental benefits of a medium-scale rooftop PV installation were evaluated. The system’s configuration, energy yield, performance ratio, and loss mechanisms were analyzed to determine operational efficiency. Economic analysis assessed potential savings, while environmental evaluation considered reductions in carbon emissions and contributions to sustainable energy goals. The results demonstrate that grid-tied PV systems can provide reliable electricity, support climate action, and offer economic advantages for academic institutions.
Turkey has not been indifferent to these developments in the world and has benefited from these positive externalities by supporting the manufacturing sector with various instruments. One of these positive externalities is employment growth. Positive effects in technology and production methods have not completely eliminated the risks and hazards that employees face at work. Therefore, occupational health and safety is an issue that needs to be addressed meticulously by all actors in working life. The public, employers, employees, non-governmental organizations and trade unions in working life and local authorities have taken important steps on the subject. However, occupational accidents continue to occur for various reasons. In this study, occupational accidents resulting in injury of manufacturing sector employees in Amasya Province were examined. By focusing on the accidents in the metal sector and garment sector, which are the sub-branches of the manufacturing sector where most occupational accidents occur, it is aimed to help the occupational health and safety measures to be taken by employers, insured employees and other stakeholders operating in Amasya Province and thus the city economy.
Recent advances in speech-based artificial intelligence have increased interest in extracting demographic information, such as gender, directly from speech signals. However, most of the existing studies have not addressed in detail the relationship between the emotional content of speech and gender prediction performance. This situation has created a gap in the literature. In this study, a deep learning-based gender recognition method was developed to fill this gap, and the results of the obtained model were analyzed in three different categories: polarity (positive, negative, neutral), emotion type (joy, fear, surprise, sadness, disgust, anger, neutral), and emotional intensity (normal, strong). The gender recognition performance was 96.18% in terms of accuracy. In addition to high performance, the model's analysis results show that gender prediction accuracy varies depending on the emotional content and intensity of the speech.
For industrial enterprises, effective energy trading in intraday markets is vital for cost reduction but challenging due to market volatility and the complexity of decision-making. This study proposes a novel integrated energy trading framework utilizing Deep Reinforcement Learning (DRL) to automate and optimize trading strategies in the Intraday Market (IDM). Unlike conventional rule-based or purely simulation-driven RL models, the proposed framework employs a state-augmented Markov Decision Process (MDP) formulation that integrates exogenous System Marginal Price (SMP) interval forecasts directly into the agent’s observation space. Specifically, a Deep Q-Network (DQN) architecture augmented with Prioritized Experience Replay (PER) is developed to handle the high volatility and continuous state space of energy markets, overcoming the limitations of traditional tabular Q-learning. To enhance the agent's decision-making capabilities, a hybrid forecasting mechanism employing LightGBM and CatBoost algorithms is integrated to predict SMP. The proposed system was validated using real-world data from Energy Exchange Istanbul (EXIST). The results demonstrate that the DRL-based agent achieved approximately $1 savings per MWh and reduced monthly operational labor costs by over $6,000 compared to rule-based strategies, achieving a ~158% performance improvement over discrete-state Q-table baselines and proving its efficacy in minimizing energy costs and managing imbalances.
Occupational accidents pose a major threat to economic enterprises in many different industries, reducing labor productivity and causing substantial economic losses, particularly in highly industrialized regions. According to ILO data, losses due to occupational accidents in Türkiye reached to 1 trillion 51 billion TL in 2023, highlighting the need for evidence-based analytical approaches to strengthen prevention strategies. Effective occupational accident analysis is essential not only for occupational safety and health (OHS) but also for sustainable development. Recent studies increasingly leverage data-driven methods, although existing findings remain fragmented and lack methodological synthesis. Alongside traditional statistical methods, modern machine learning (ML) techniques offer enhanced analytical capabilities for accident analysis. Machine learning methods, including prediction, classification, and clustering algorithms, are increasingly used to support risk identification, pattern analysis, and data-driven safety insights by learning from historical data and modeling complex accident patterns. This study systematically reviews existing research on ML applications in occupational accident analysis, examining investigated sectors, employed algorithms, data sources, and analytical trends. The review also identifies key research gaps and outlines future directions for ML-driven OHS studies.
Recent breakthroughs in artificial intelligence have accelerated the adoption of learning algorithms, visual data categorization, and defect identification methods within quality control workflows across manufacturing industries. This review thoroughly explores the incorporation of AI-based solutions within industrial workflows, emphasizing the improved performance of neural network-driven techniques as opposed to classical inspection strategies. Specifically, CNN architectures (convolutional neural networks) and advanced YOLO models (YOLOv5, YOLOv7, YOLOv8) have exhibited outstanding precision and speed in detecting structural anomalies and identifying faults as they occur. By utilizing attention mechanisms, architectural refinements, knowledge transfer techniques and synthetic data expansion, these models have demonstrated enhanced capabilities—particularly in cases where training data is scarce. Transformer-based and hybrid neural network designs have shown great potential as effective alternatives, ensuring a balance between accuracy and computational demand. Research findings suggest that AI-based quality control frameworks play an essential role in minimizing manufacturing defects, elevating product quality, and improving overall customer satisfaction. These advancements span multiple industries, including automotive, steel, textiles, electronics, and more. They strongly align with Industry 4.0 goals, enabling the shift towards digitalized manufacturing systems, thereby fostering intelligent and automated production environments.
The development of titanium alloys in various medical fields is currently of great interest to researchers, and in this regard, the manufacture of various parts such as hip joints, knee joints, surgical equipment, and dental implants is of great interest. This study is a review of the latest achievements in this field, and the results show that the development of new generation alloys containing elements such as niobium and molybdenum can lead to the manufacture of wear-resistant alloys with lower elastic modulus, which are used in dental implants. This article also reviews some applications of titanium quantum dots, and it is suggested that researchers in the future should work on the study and development of quantum dots and their use in the development of quantum alloys.
As semiconductor technology evolves, electronic systems that operate at low power are becoming more prevalent and attract the attention of researchers. In this study, a single-phase voltage controlled inverter was designed and implemented using the H-bridge topology and Pulse Width Modulation (PWM) technique for systems requiring low power. MOSFETs and ATMEGA328P-AU microcontroller were used as switching circuit elements in order to obtain high efficiency in the design. LC filter was used to purify the output signal from harmonics and to obtain a pure sinusoidal. Circuit diagrams were created using Proteus Design Suite and Altium Designer programs. The design was turned into a device and final checks were made in the laboratory environment, and it was seen that the system worked smoothly, and the simulation studies and experimental results coincided. The designed product gives a stable AC voltage with 220V peak voltage and 50Hz frequency.
Early diagnosis and precise detection of skin cancer represent a global health priority since this disease remains highly dangerous while being among the most frequent ones. This research investigates the effectiveness of deep learning techniques, specifically Convolutional Neural Networks (CNN) and the VGG16 architecture, for skin cancer detection and classification. The study works with images from the International Skin Imaging Collaboration (ISIC) while employing resizing and augmentation preprocessing to boost its model performance. We evaluate the proposed model using precision, recall, and F1-score metrics to ensure accurate classification. The proposed CNN model achieved 87% validation accuracy, outperforming the VGG16 model, which attained 65% accuracy. Experimental results highlight the potential of AI-driven models in improving diagnostic accuracy, demonstrating their significance in medical image analysis and early skin cancer detection.
In this study, the catalytic activity of starch, cellulose and coffee were investigated in the dehydrogenation of sodium borohydride. The hydrogen generation rates of starch, cellulose and coffee were measured as 4.0, 6.7 and 60 ml H2 min-1 g-1 and the activation energies of the reactions were calculated as 27.4, 17.1 and 14.5 kJ mol-1 for starch, cellulose and coffee respectively. The study showed that natural sources could be used directly as catalysts in the dehydrogenation of chemical hdyrides.
The straight heatsink is one of the most common heat transfer components used in desktop CPUs to manage the heat generated by the microprocessor. The study aimed to find the optimal fin numbers of the straight heatsink for three different fin thicknesses and compare the masses at these points. For the analysis, the present study used Solidworks® software to create CAD models and perform the CFD simulation. It was found that each of the three different fin thicknesses had a turning point at which the microprocessor’s temperature was at its minimum. The weight of the heatsink was also measured at those turning points. Specifically, the heatsinks with 1 millimeter, 1.5 millimeters, and 2 millimeters thickness had a microprocessor temperature of about 83.52 degrees Celsius, 86.50 degrees Celsius, and 89.25 degrees Celsius, with the weight of approximately 307.80 grams, 388.80 grams, and 448.2 grams. Overall, a 1-millimeter fin thickness with 21 fins configuration for this study was best under the criteria of minimum microprocessor temperature and minimum heatsink mass. Thus, this study successfully demonstrated that optimization of mass and fin thickness of the heatsink was possible to provide better thermal management of the microprocessors of a desktop’s CPU. This study is significant for this era because it provides a panacea for minimum material cost, lightweight, and minimum microprocessor temperature.
Underwater wireless optical communication systems face significant challenges due to the heterogeneous nature of the underwater environment and the attenuation of optical signals caused by absorption and scattering. These effects restrict the data transfer capacity and transmission distance, resulting in communication errors. Different modulation techniques are used to minimize the effects of these parameters. Automatic modulation classification plays a critical role in terms of effective management of spectrum resources. In this study, underwater wireless optical communication channels are modulated with different modulation techniques, and the signals are transformed into the discrete wavelet space, resulting in approximation and detail coefficients that are used as feature vectors for training machine learning algorithms. In addition, optimized classification features are determined for different signal-to-noise ratios and different transmission distances using the genetic algorithm. The results show that the approximation and detail coefficient energies provide higher classification performance in the classification of modulated signals according to statistical features such as mean, variance, and standard deviation. According to simulation results, an average classification accuracy of 82% has been obtained using the proposed discrete wavelet transform and genetic algorithm-based technique, which demonstrates high classification accuracy for noisy underwater channels.
Usability analysis in software has a critical role in creating useful products for users and obtaining the necessary feedback. Learning Management Systems (LMSs) are among the most frequently used software in distance education. The usability of these systems is crucial for both student success and the quality of the service offered. This study examines LMS usability through faculty perceptions at a state university, employing a mixed-methods design. Quantitative data were collected using the System Usability Scale (SUS) from 109 faculty members, while qualitative insights were gathered from semi-structured interviews with nine participants. Statistical and thematic content analyses were employed to interpret and compare results. The quantitative analysis yielded an average SUS score of 63.85 ± 16, indicating moderate usability concerns. Based on findings from both the SUS responses and interview data, several recommendations were proposed, such as enhancing system infrastructure, simplifying the interface, improving instructional guidance, strengthening interaction features, optimizing file management, refining notification systems, addressing character encoding issues, and streamlining listing and reporting functions. The results underscore the importance of a user-centered development approach, incorporating participatory design principles. Future research should track how faculty adapt to LMS updates over time. We hope these findings will guide future usability studies.
