University of Ulsan Open Access Korea
Not a member yet
17799 research outputs found
Sort by
수소 생성 반응 효율 최적화를 위한 나노촉매의 전략적 설계 및 평가
No full textEnvironmental pollution caused by the use of fossil fuels and the depletion of resources are critical challenges facing modern society, making the development of sustainable energy sources essential. Among these, hydrogen energy is gaining increasing importance as a clean and renewable energy source, with significant research efforts directed towards improving the efficiency of hydrogen evolution reactions (HER). Although Pt is well known for its excellent catalytic performance in HER, its high cost and limited availability hinder its widespread commercialization. This study addresses these limitations by developing high-performance and cost-effective catalysts. The strategy of this research was to optimize HER efficiency by controlling the shape and alloying of the catalysts. We synthesized porous Pt2NiCo nanosheet catalysts and applied them to electrochemical HER under alkaline conditions. Additionally, Pt and Ni were doped onto the Re(0001) surface to form Pt-Ni@Re(0001) NPCs, and their effects on electrochemical activity under alkaline conditions were analyzed using density functional theory (DFT) calculations. The results demonstrated that Pt2NiCo nanosheets exhibit superior HER activity and stability compared to Pt3Ni, Pt3Co, Pt nanosheets, and commercial Pt/C. Furthermore, Pt-Ni@Re(0001) NPCs were found to play a crucial role in enhancing the electrochemical activity for hydrogen evolution by altering the electronic structure of Re through Pt and Ni doping. This study proposes a catalyst design strategy that mitigates the limitations of Pt catalysts while improving the cost-efficiency of hydrogen energy, thus making significant contributions to the development of renewable energy technologies.Maste
An investigation of the operating conditions to improve spark ignition engine’s efficiency and emission characteristics
No full textAn investigation of the operating conditions to improve spark ignition engine’s efficiency and emission characteristics Department of Mechanical Engineering Quach Nhu Y This dissertation presents a thorough and systematic investigation into the optimization of key operating parameters to improve thermal efficiency and reduce emissions in small SI engines. Recognizing the limitations of conventional experimental approaches such as high cost, restricted flexibility, and difficulty in accurately quantifying residual gases. This research adopts a hybrid methodology that combines high-fidelity experimental testing with advanced one-dimensional simulations using AVL-Boost software. The experimental setup was designed to capture essential engine performance data under full-load, steady-state conditions, including ignition timing, air mass flow, in-cylinder pressure and temperature, and engine torque. These data were then used to calibrate and validate a detailed simulation model capable of predicting engine behavior and emissions across a wide range of operating conditions. The dissertation is structured into nine chapters. Chapter 1 introduces the background, research objectives, and scope of the study. Chapter 2 provides a comprehensive literature review, identifying research gaps related to combustion duration, ignition timing, fuel heating value, residual gas estimation, and bore-to-stroke ratio effects in spark-ignition engines. Chapter 3 details the experimental setup, including the small engine test platform and measurement techniques. Chapter 4 explains the development and validation of the AVL-Boost simulation model based on experimental data. Chapter 5 presents the ignition timing analysis. Chapter 6 focuses on combustion duration, identifying an optimal range that enhances energy release while minimizing heat loss, thereby improving both efficiency and emissions. Chapter 7 explores the impact of fuel heating value on combustion characteristics. Chapter 8 focuses on predicting multiple engine performance and emission results at the same time using deep neural networks (DNNs) for propane spark-ignition engines. It explains how the model was built, trained, and tested. Finally, Chapter 9 summarizes the key findings, highlights the main contributions of the research, and discusses its implications for improving small-engine performance through integrated experimental, simulation, and machine learning approaches. Overall, this research provides a practical and accurate approach to improving small spark- ignition engine performance and emissions, offering valuable guidance for future development of cleaner and more efficient engine technologies. The results show that advancing the spark improves engine efficiency and BMEP but leads to increased NOx emissions, while retarding timing reduces emissions at the cost of performance. Moreover, higher energy content contributes to better performance. In addition, the model can make accurate predictions under different engine conditions, with strong agreement between predicted and actual values. This shows the model is useful for helping improve engine performance quickly and effectively.Docto
Development of a Machine Learning-Based Diagnostic Tool for Long COVID
No full textCOVID-19 세계적 대유행은 종료되었지만, 감염 이후 4주 이상 지속되는 다양한 증상인 long COVID는 여전히 많은 환자에게 영향을 미치고 있다. 이 중 일부는 1년 이 상 증상이 지속되어 삶의 질 저하를 초래한다. Long COVID는 진단 기준이 명확하지 않 으며 병태생리와 바이오마커가 확립되지 않아 임상적 판단이 어렵다. 또한 long COVID 를 평가하는 기존의 설문조사는 문항 수가 많고 무응답률이 높다. 이에 본 연구는 머신 러닝을 활용하여 후유증 예측 핵심 문항을 도출함으로써 평가도구의 간소화를 목표로 하였다. 2022년 11월부터 2024년 9월까지 수집된 COVID-19 감염자 및 비감염자의 자가 보고 설문 응답지를 분석에 활용하였다. 총 290명으로부터 수집된 설문 중 응답률이 30% 미 만으로 응답을 보인 참여자를 제외한 276명의 693건의 설문 자료를 최종 분석에 사용 하였다. COVID-19 감염자는 확진일을 기준으로 1개월, 3개월, 6개월, 12개월에 시행 된 설문 자료를 분석에 포함하였다. 비감염자는 COVID-19에 감염된 적이 없다고 진술 한 대상에서 수집된 설문을 포함하였다. 머신러닝 모델 분석 결과 불안과 우울(HADS) 항목의 13번, 4번, 6번, 9번, 8번 문 항, 피로(FAS) 항목의 4번, 1번, 10번 문항, 그리고 COVID-19 후유증 증상 중 근력저 하, 피로 문항이 상위 10개 주요 예측 문항으로 도출되었다. 10개로 간소화한 설문 문 항을 사용한 경우 ROC-AUC는 0.993, Precision-Recall 곡선의 AUC는 0.994로 나타났 다. 감염 여부를 구분할 수 있는 임곗값(threshold)은 0.528로 도출되어 감염자 81명 중 78명이 감염자로 분류되었고 비감염자 78명은 모두 비감염자로 분류되었다. 간소화 된 10개 문항 기반 long COVID 예측값은 삶의 질 지표인 EQ-5D-5L (r=.358, p<.001) 및 EQ-VAS (r=.361, p<.001)와 유의한 양의 상관관계가 확인되었다. 간소화 된 진단 도구는 long COVID 선별을 위한 실용적인 기준으로 활용 될 수 있음 을 시사한다. 간소화된 문항을 사용하면 삶의 질 저하와 밀접한 증상으로 구성되어 있 어 감염 후 신체 및 정신 기능 저하를 조기에 파악하는 데 도움이 될 것으로 사료된 다. 본 연구 결과는 향후 롱코비드 정의와 진단 기준 개발을 위한 기초 자료로 활용될 수 있다. 주요어: COVID-19감염 후 후유증, 머신러닝, 설문, 간소화, 삶의 질Maste
Energy-Efficient Blockchain-Enabled IoT Networks:A Deep Reinforcement Learning Approach
No full textEnergy-Efficient Blockchain-Enabled IoT Networks:A Deep Reinforcement Learning Approach ZERIHUN HURUY NEGASH Supervised By: Professor Sungoh Kwon Submitted in Partial Fulfillment of the Requirements for the degree of Master of Science The growing adoption of Internet of Things (IoT) calls for secure, scalable, and energy-efficient blockchain solutions. However, conventional blockchains are often too energy-intensive and slow to meet IoT demands. In this study, we propose a reputation- based reliable, available, fault-tolerant (R-RAFT) protocol that specifically addresses the need for optimal leader election and block size management, thereby enhancing the overall performance of blockchain networks in supporting IoT applications. Our solution focuses on reducing energy consumption by optimizing key processes—leader election, and dy- namic block size generation. Each of these factors directly impacts the energy use, block commitment time, and service satisfaction levels in IoT networks. To achieve these op- timizations, we employ a Deep Reinforcement Learning (DRL)-based approach that dy- namically manages communication resources, adapts block sizes, schedules IoT data com- mitments efficiently, and non-linearly maps reputation to timeout durations. This joint optimization approach ensures that energy consumption is minimized while consistently maintaining data commitment times and service satisfaction within target limits. Simula- tion results validate that our blockchain framework for wireless communication effectively improves the overall efficiency and adaptability of blockchain-enabled IoT networks. Keywords: Internet of Things (IoT), Blockchain, Block Commitment Time, Dynamic Block Size, Communication Resources, Deep Reinforcement Learning,Maste
Development of AI-based video analysis model for assessing cognition and grasp pattern in pediatrics
No full textThis paper presents two AI-based video analysis approaches for the quantitative assessment of developmental status in infants and young children. The first study aimed to develop and validate an AI-assisted evaluation tool for infants aged 12 to 42 months to assess cognitive function. A total of 75 participants, referred for developmental assessment or suspected of developmental delay, were evaluated using the Bayley Scales of Infant Development II (BSID-II). Video recordings of the “Places Pegs in” and “Blue Board” tasks were automatically analyzed using a YOLOv5-based AI model and compared with clinician-administered BSID-II results. The AI tool achieved 86.5% sensitivity and 100% specificity in the “Places Pegs in” task, and 96.9% sensitivity and 89.5% specificity in the “Blue Board” task. The second study focused on classifying grasp pattern during the “Grasp of Cereal” task from the Quality of Upper Extremity Skills Test (QUEST) in 8 typically developing children, 9 children with developmental delay (mean corrected age 28.24 months) and 65 children with cerebral palsy (mean age approximately 5 years). A 3D Convolutional Neural Network (3D CNN) and Segment Anything Model 2 (SAM2) were used to analyze temporal sequences of children reaching for a “Cereal” object, effectively capturing spatiotemporal characteristics of hand motion. Together, these studies demonstrate the potential of AI-based video analysis tools as accurate and practical support systems for developmental assessment and diagnosis in pediatric clinical settings.Maste
A Study on Automatic Landmarks detection of cephalogram, Skeletal pattern classification, Surgical and Growth movement prediction and Generation of post-operational lateral cephalogram
No full textOrthognathic surgery is essential for enhancing both aesthetics and function in patients with severe skeletal malocclusions. However, planning these surgeries requires labor-intensive manual processes, which include cephalometric landmark selection, cephalometric analysis, and simulation of surgical movements of bones. This study aimed to address these challenges by developing AI-driven solutions to improve efficiency and accuracy
in these tasks. We first created a deep learning-based model for automatic landmark selection, reducing the manual workload and streamlining the diagnostic process. Additionally, we developed a one-step diagnostic model capable of identifying anteroposterior skeletal discrepancies (APSDs) directly from lateral cephalograms, eliminating the need for manual measurements and improving workflow efficiency. We also introduced a surgical movement prediction model that utilizes preoperative cephalograms and detected landmarks to estimate surgical movement, aiding in surgical planning. Furthermore, this model was extended to growth prediction in children aged 8 to 10 years, supporting growth-related treatment planning. Finally, a conditional latent diffusion
model was developed to simulate postoperative skeletal and soft tissue changes. This digital twin framework offers personalized treatment planning by visualizing potential surgical outcomes. By automating key processes, this study demonstrates the potential of AI to enhance the efficiency for diagnosis and treatment planning in orthodontics and orthognathic surgery, ultimately supporting improved patient outcomes.Docto
Development of Heterogenous Polymeric Chemosensor for the Recognition and Removal of Environmentally Toxic Pollutants
No full textChapter 2: 2,4,6-Trinitrophenol, widely known as picric acid, is a highly explosive compound historically used in military-grade ammunition. Due to its extreme toxicity and high sensitivity to heat and friction, it poses a significant risk of accidental detonation at elevated temperatures. Therefore, prompt detection and isolation of picric acid in affected areas is essential. In this study, we developed a fluorescent copolymer (P1) specifically designed to detect picric acid under elevated temperature conditions. The polymer's solvatochromic behavior and aggregation-induced emission (AIE) properties enabled a fluorescence "turn-off" response when exposed to picric acid at 40 °C—a critical temperature threshold. To enhance usability, we fabricated a thin polymer film (F1) by anchoring P1 onto a quartz slide. Remarkably, F1 maintained its luminescent behavior over a wide temperature range, thanks to its AIE characteristics. This portable film demonstrated exceptional sensitivity to picric acid at nanomolar concentrations in neutral aqueous environments, making it suitable for field applications. Moreover, F1 effectively separated picric acid from mixtures of nitroaromatic explosives with a 70% efficiency across four reuse cycles. Its success in real-sample testing underscores F1's promise for environmental monitoring and practical deployment in contamination scenarios.
Chapter 3: Paraquat herbicide poses a significant threat to human health, food safety, and environmental sustainability. Despite its widespread use, the development of rapid, selective, and sensitive methods for on-site detection and removal remains a major challenge. In this study, we introduce a fluorescent polymeric probe (P1) for the detection of paraquat via a fluorescence "turn-off" mechanism based on photoinduced electron transfer (PET). Probe P1 was synthesized through free radical polymerization and consists of three key components: hydrophilic monomer N,N′-dimethylacrylamide, the paraquat recognition unit pyren-1- ylmethyl methacrylate, and the substrate cross-linking unit benzophenone acrylamide. P1 exhibited outstanding selectivity for paraquat, effectively distinguishing it from structurally similar herbicides and metal ions commonly found in real environmental samples. When fabricated into a polymeric thin film (F1) on a glass substrate, the system demonstrated enhanced sensitivity, achieving a low detection limit of 0.011 mM and a high Stern–Volmer constant of 1.05 × 10⁴ M⁻¹. The portable F1 device enabled reliable and accurate detection of paraquat across a variety of real-world samples, including water sources, soil, apples, and lettuce. In addition to its superior sensing performance, F1 also exhibited 100% removal efficiency of paraquat from a complex mixture of herbicides. Notably, the film retained its sensing and removal capabilities over four consecutive reuse cycles, underscoring its practicality, cost-effectiveness, and strong potential for deployment in large-scale environmental monitoring and remediation.
Chapter 4: Polychlorinated biphenyls (PCBs) are highly toxic, hydrophobic pollutants that resist degradation and persist in ecosystems, posing risks to human health and the environment. Early detection, removal, and monitoring are therefore essential. This study developed a novel fluorescent polymer probe, P1, synthesized from N,N'-dimethylacrylamide (DMAA) and pyren-1-ylmethyl methacrylate (PyMMA), to detect PCB congeners 77, 118, and 126 in water. P1 showed high sensitivity to the toxic coplanar PCBs 77 and 126 via a fluorescence turn-on mechanism driven by hydrophobic and π–π interactions. Detection limits were 0.028 mM for PCB 77 and 0.039 mM for PCB 126, while PCB 118 exhibited lower sensitivity. Additionally, a porous 3D polymeric organogel (OG) composed of butyl acrylate, PyMMA, and ethylene glycol dimethacrylate was synthesized for practical use. The OG selectively removed PCBs with removal efficiencies of 63% (PCB 77) and 55% (PCB 126), benefiting from its large surface area and enhanced hydrophobic and π–π interactions. This study highlights P1’s effectiveness for PCB detection and OG’s potential for PCB removal, providing a promising approach for environmental monitoring and remediation.Docto
Research on the Development Strategies of China’s Tourism Service Trade
No full text중국 관광서비스무역 발전방안 연구 ZHUANG YUMING 울산대학교대학원 정치외교학과 국제 무역에서 서비스 무역의 영향력이 지속적으로 확대됨에 따라, 관광 서비스 무역이 서비스 무역에서 차지하는 비중 또한 꾸준히 증가하고 있다. 각국은 관광 서비스 무역 발전에 점점 더 많은 관심을 기울이고 있으며, 중국 역시 예외가 아니 다. 지난 10 여 년간, 중국의 관광 서비스 무역은 빠르게 성장하였고, 관광 산업의 규모는 지속적으로 확대되었으며, 10 년 전과 비교하여 다양한 방면에서 큰 발전을 이루었다. 중국은 관광 서비스 무역과 관련된 여러 정책을 제정하였고, 동시에 다른 국가들 과의 협력을 적극적으로 추진하여 관광 경제를 활성화하였다. 이를 통해 중국의 관 광 서비스 무역 수준은 새로운 높이에 도달하였다. 그러나 이러한 지속적인 발전에 도 불구하고, 중국의 관광 서비스 무역은 여전히 발전의 여지가 많다. 예를 들어, 관광 서비스 무역의 수출입 규모는 더욱 확대될 필요가 있으며, 국제 경쟁력 또한 강화해야 한다. 중국 관광 서비스 무역의 영향 요인에 대해 여러 학자가 다양한 연구를 진행했지 만, 여전히 일정한 한계를 가지고 있다. 중국의 관광 서비스 무역의 강점을 지속적 으로 유지하고 새로운 단계로 도약시키며, 이를 중국의 핵심 산업으로 발전시키기 위해 본 연구는 현재 중국 관광 서비스 무역의 현황을 분석하고, 관광 서비스 무역 의 이론적 영향 요인을 검토하였다. 연구 결과는 중국 관광 서비스 무역의 발전에 중요한 시사점을 제공할 것이다. 본 연구에서는 우선 관광 서비스 무역의 발전 현황을 살펴보고, 최근 15 년간 중 국 관광 서비스 무역의 수출입 규모, 출입국 상황, 무역 구조, 시장 점유율, 무역 개 방도 등을 분석하였다. 연구 결과 경제 발전 수준, 환율 변동, 관련 산업 발전, 물가 변동 등이 관광 서비스 무역에 영향을 미치는 것으로 나타났다. 마지막으로 실증 분석 결과를 바탕으로 중국 관광 서비스 무역에 대한 제언을 제시하였으며, 여기에 는 경제적 요인과 비경제적 요인에 대한 구체적인 제언이 포함되어 있다. 키워드:관광 서비스 무역;국제 경쟁력;무역 발전 전략Maste
Numerical Study on Ionic Liquid-Piston Compressors for Hydrogen Refueling Stations
No full textThis study presents a detailed numerical investigation of a liquid-piston compressor employing hydrogen gas and the ionic liquid 1-ethyl-3-methylimidazolium tetrafluoroborate, using three-dimensional computational fluid dynamics simulations performed in ANSYS Fluent. The study focuses on the compression and expansion processes, offering comprehensive insight into fluid flow, heat transfer mechanisms, and the thermodynamic performance of liquid-piston compressors under varying operational and geometric conditions. A two-phase simulation framework using the finite volume method and volume of fluid technique was implemented to accurately track the gas-liquid interface and resolve governing equations for mass, momentum, and energy conservation. Experimental validation of hydrogen pressure trends confirmed the reliability of the model. During the compression phase, 1-ethyl-3-methylimidazolium tetrafluoroborate was injected into a cylindrical chamber, compressing hydrogen from 220 bar to 752.3 bar, achieving a compression ratio of 3.4. The dynamic behaviors of pressure and temperature were segmented into three distinct stages characterized by linear or quadratic trends. The effects of liquid velocity, initial chamber pressure, and working fluid type were systematically explored. Results indicated non-linearities in hydrogen pressures with changes in inlet velocity and initial pressure. While 1-ethyl-3-methylimidazolium tetrafluoroborate and 1- butyl-3-methylimidazolium bis(trifluoromethanesulfonyl)imide showed comparable performance, increasing the surface area-to-volume improved heat dissipation, reduced peak temperatures, and significantly influenced the time required to reach target pressures. To enhance thermal performance, the liquid-piston compressor was geometrically modified with internal cooling pipes, where simulations revealed that circular cross-sections yielded the best cooling performance. Cooling effectiveness improved as pipe diameter and the number of pipes increased, with maximum reductions in hydrogen temperature reaching 39.2 K. These modifications enabled compliance with the U. S. Department of Energy guidelines for hydrogen storage, keeping gas temperatures below 358 K while boosting compression efficiency to 95.8%. In exploring non-constant inlet velocities‒such as sine, rectangular, triangular, and sawtooth waveforms‒the study found that although compression efficiency remained relatively stable, thermal performance could be improved. Notably, wave amplitude and waveform type directly impacted the thermodynamic state of hydrogen, with triangular and sawtooth profiles causing distinctive oscillatory behaviors in pressure and temperature. Moreover, variable flow strategies produced lower hydrogen temperatures at target pressures compared to constant velocity conditions, indicating potential benefits for thermal management. The expansion phase was also modeled to examine the reversibility of the system and energy recovery capabilities. Hydrogen was initially at 859.4 bar and 405.4 K, and after expansion through 1-ethyl-3-methylimidazolium tetrafluoroborate outflow, it reached 196.7 bar and 266.3 K, demonstrating effective decompression. The study revealed that wall temperature, liquid temperature, and discharge velocity significantly impacted expansion efficiency and power density. The pressure and temperature followed predictable power-law and log-linear relationships with time and volume, suggesting that the expansion process could be approximated using simplified thermodynamic models. Overall, this research provides critical insight into the fluid dynamics and thermal behavior of liquid-piston compressors and lays a solid foundation for their optimization in hydrogen compression applications. The findings serve to guide the design of energy-efficient and thermally stable hydrogen storage systems aligned with future clean energy infrastructure.Docto
RESEARCH ON RSMA WITH HYBRID STAR-RIS FOR COVERT COMMUNICATION
No full textCovert communication poses a substantial challenge in wireless communication, as it can hinder the effectiveness of protecting sensitive messages and improving overall system performance. In this work, we propose an effective approach to enhance the robust covertness of hybrid active-passive simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS)-based rate splitting multiple access (RSMA) systems, which require a specific level of security. Under Rician fading channels, we initiate our analysis by deriving the closed-form expression for the detection error probability (DEP) linked to an eavesdropper monitoring the covert information of nearby users, then establish the optimal DEP for the worst-case scenario, where it serves as a critical parameter in the system’s security measures. Afterward, the non-convex optimization problem of determining the power allocation (PA) resources is addressed to maximize the covert rate by reformulating it into two manageable sub-optimization problems. We leverage the analysis of the derived DEP to obtain closed-form solutions for the PA coefficients in each sub-problem. These solutions are then addressed using an iterative algorithm characterized by relatively low computational complexity, ensuring efficient resolution of the optimization tasks. Numerical results validate the precision of our analysis and demonstrate the effectiveness of our proposed optimization solution. In addition, several key findings emerge from our study. Notably, the RSMA system outperforms NOMA in terms of security efficiency. Furthermore, utilizing a hybrid mode allows for flexible adjustment of the number of STAR-RIS elements, optimizing performance by employing active mode at high transmission power and passive mode at lower power levels.Maste