Urban traffic congestion continues to intensify owing to rapid urbanization and growing vehicle ownership, highlighting the limitations of fixed-time signal control systems. This paper proposes a real-time traffic signal optimization framework that integrates drone-based object detection and tracking with a genetic algorithm and rolling horizon structure. Traffic data were collected at the Hakha intersection in Daejeon, South Korea, using DJI M300 RTK and Mavic 3E drones during the evening peak hours. Vehicle detection and tracking were performed using YOLOv8n and ByteTrack, achieving an average detection accuracy of 88–98% for the total approach volume and 84–94% for the through-movement volume. The extracted traffic parameters (volume, delay, and queue length) were incorporated into a multi-objective fitness function with weights determined via the analytic hierarchy process. The optimized signal plans were validated using VISSIM microsimulation against a fixed-time baseline. Results show that the proposed framework reduced the average vehicle delay by 31.4% (37.91 to 25.99 sec/veh), stop delay by 37.1%, and average queue length by 34.8% (8.41 to 5.48 m), while improving the intersection level of service from D to C without sacrificing network throughput. This study demonstrated the practical feasibility of an integrated framework combining drone-based mobile sensing and metaheuristic optimization for real-time adaptive signal control.

본 연구는 막 손상을 검지하기 위한 압력손실시험(pressure decay test, PDT)에서 분리막의 초기 상태(virgin state) 의 물성을 고려하여 산정하는 초기설정압력(Ptest, Pmax)에 대한 오차범위를 확인하고자 하였다. 여과 공정을 수행하면서 불가 피하게 발생하는 막오염으로 인하여 변화하는 접촉각을 연속적으로 측정하여 막간차압(transmembrane pressure, TMP)과의 상관관계를 확인한 결과, 막간차압이 0~200 kPa 증가할 때 막 표면에서의 접촉각이 83~41° 범위로 선형 감소하는 결과로 측 정되었다. 물리세척에 의한 회복율이 높은 가역적인 오염구간보다 비가역적인 막오염 구간에서 접촉각의 변화폭이 커지는 것 을 확인할 수 있었다. 이를 통하여 종래 고정된 초기설정압력에서 막의 물성과 막오염도에 따라 적용의 필요성을 확인할 수 있었으며, 막 손상 면적 대비 일정한 압력손실율(pressure decay rate, PDR)을 얻기 위해서 압력손실시험을 수행할 때 막오염의 간접지표인 TMP를 고려하여 동적으로 초기설정압력값을 산정 및 적용해야만 절대적인 수질 보증을 유지할 수 있을 것이다.

Rapid, real-time detection of anomalies and locate structural defects during earthquakes is critical for ensuring safety and enabling timely decision-making. Although deep learning-based structural health monitoring (SHM) has shown considerable promise, conventional supervised models are often impractical because labeled damage data from real-world structures are extremely scarce. To address this challenge, this paper proposes a Multi-Class Deep Support Vector Data Description (SVDD) framework for structural defect detection. The proposed Multi-Class Deep SVDD approach learns the boundary of normal data using only normal seismic acceleration responses. When new data are recorded, the system infers both the occurrence and location of defects by evaluating whether the responses fall within or deviate from the learned normal boundary. The framework is validated using the Los Alamos National Laboratory 3-story bookshelf structure benchmark dataset. Experimental results show that the proposed model achieves a peak average accuracy of 87.12% in a 4-dimensional latent space, substantially outperforming traditional baseline methods, including Kernel Density Estimation (KDE), SVDD, and One-Class Deep SVDD. These findings indicate that the Multi-Class Deep SVDD framework provides a robust and objective metric for rapid post-earthquake safety assessment without requiring prior exposure to faulty datasets.

Plastic injection molding is widely used in the automotive, electronics, and other manufacturing industries. Since product quality is significantly affected by process variables such as temperature, pressure, speed, and cycle time, datadriven quality management has become increasingly important. However, conventional quality control methods mainly depend on operator experience and post-process inspection, making it difficult to detect defects and abnormal process conditions in real time. In addition, manufacturing datasets often suffer from severe class imbalance because defective samples are much fewer than normal samples. This study proposes an AI-based defect prediction and anomaly detection system using plastic injection molding process data. For defect prediction, XGBoost was selected as the supervised learning model, and SMOTE was applied to address class imbalance. Recall, F1-score, and PR-AUC were used as the main evaluation metrics instead of accuracy. SHAP analysis was also applied to identify important process variables affecting defect occurrence. For anomaly detection, Isolation Forest and AutoEncoder were used together. Isolation Forest was adopted for fast first-stage detection, while AutoEncoder was used to detect complex nonlinear abnormal patterns. The results indicate that the proposed system can support both defect prediction and process anomaly detection. In the supervised learning results, XGBoost showed the best performance on the CN7 dataset, achieving a recall of 1.000 and an F1-score of 0.800. However, all supervised models showed limited defect detection performance on the RG3 dataset, indicating the effect of severe class imbalance and dataset-specific characteristics. In anomaly detection, AutoEncoder achieved the highest F1-score among the compared models, while Isolation Forest demonstrated practical advantages for real-time application due to its computational efficiency. Therefore, the proposed AI-based system can contribute to smart quality management and data-driven decision-making in plastic injection molding processes.

Sensing toxic gas molecules is crucial for environmental monitoring and human safety. In this spin-polarized DFT study, we introduce BCN nanocage, a hybrid analogue of C₂₀ nanocage, for sensing Cl2, COCl2, H2S, and NH3 gas molecules. BCN is functionalized with a series of metal adatoms (Li, Be, Al, Si, P, Sc, Ti, V, Mn, Fe, Ni, and Cu), chosen for their diverse electronic configurations and potential to interact strongly with the substrate. Among these, only the Li-, Al-, Sc-, Fe-, and Cu-decorated BCN complexes were found to be thermodynamically stable and energetically favourable. Sc exhibits the strongest binding with the nanocage, followed by Fe, Al, Li, and Cu, due to bond formation and significant charge transfer from adatoms to the nanocage. Among the studied candidates, BCNCu emerges as the most promising for Cl2 sensing under dry conditions, exhibiting an adsorption energy of 0.66 eV, a recovery time of 0.02s, and a -37.16% band gap variation. Compared with previously reported nanocage-based sensors, BCNCu demonstrates a balanced combination of suitable adsorption energy, rapid recovery time, and appreciable sensitivity, highlighting its potential for efficient Cl2 detection under dry conditions. However, its sensing performance is influenced by humidity, indicating that BCNCu operates more effectively under dry conditions than in humid atmospheres. AIMD simulations and vibrational spectra analysis confirm the thermal stability of the substrate at 400 K and its dynamical stability. This study advances the field by establishing BCNCu as a promising Cl2 sensor while highlighting its limitations in humid environments, offering valuable insights for experimental fabrication and real-world applications.

Injection-molded products frequently exhibit localized surface defects such as weld lines, flow marks, scratches, bubbles, and burn marks due to variations in material flow, mold temperature, and cooling conditions. Conventional visual inspection is highly dependent on operator experience, while rule-based machine vision methods are limited under variations in lighting and surface texture. This study proposes a deep learning–based defect detection model using YOLOv8 combined with a novel Defect-Aware Augmentation technique designed to enhance robustness for small, local defect regions. The proposed augmentation pipeline includes geometric transformations, optical perturbations, local defect patch synthesis, and diffusion-based synthetic defect generation. Experiments were conducted on a custom dataset of 5,000 images (3,000 normal and 2,000 defective). Results show that the proposed model achieves significant improvements over baseline models, obtaining 95% precision, 90% recall, and 0.96 mAP@0.5, outperforming the default YOLOv8 model by 7%p in mAP. Ablation studies verify that defect-aware augmentation is the dominant factor contributing to the performance gain. The proposed system demonstrates high applicability for automated quality inspection in injectionmolding production lines.

This study evaluated the effects of image preprocessing techniques on detection of Fire Department Connection (FDC) in road view images using a YOLOv8s-based framework. Six preprocessing techniques were applied under identical training and evaluation settings, and their performances were assessed using precision, recall, mAP@50, and mAP@50–95. Geometric correction produced the largest improvement, increasing mAP@50–95 from 0.419 to 0.543 and also improving recall, indicating enhanced localization and detection stability. HSV (Hue, Saturation, Value)-based red restoration achieved the highest average precision among color-based methods, whereas Retinex-based illumination correction degraded the performance across all metrics. Bottom-region cropping improved localization accuracy but reduced recall owing to limited spatial coverage. These results demonstrate that distortion mitigation and selective color enhancement are effective preprocessing strategies for robust FDC detection in road view environments. The study provides practical guidelines for intelligent road asset management, contributing to optimized road network operation and accessibility by reducing emergency vehicle positioning time in complex urban road environments.

노후화된 사회 기반 시설물 증가에 따라 정기적인 구조물 손상 점검의 중요성이 확대되고 있다. 그러나 기존 점검 방식은 고가의 장비와 다수의 인력을 요구하며, 차선 폐쇄를 필수적으로 수반한다. 특히 차선 폐쇄는 교통 체증을 유발해 차량의 반복적인 가속과 감속, 공회전을 증가시키고 결과적으로 연료 소비와 온실가스 배출량을 증가시켜 사회적 비용을 초래한다. 이에 AI 기술을 활용해 차선 폐쇄 없이 손상을 탐지하는 연구가 진행되고 있으나 대부분 도로포장 탐지에 한정되어 있어 교량 기둥이나 방호 울타리 등 입체 구조물에 대한 탐지 기술과 차선 폐쇄에 따른 운영 효율성 및 에너지와 배출량 변화에 대한 정량적 분석은 부족한 실정이다. 본 연구는 차선 폐쇄 없이 사회 기반 시설물의 손상을 탐지할 수 있는 AI 기반 손상 시스템을 구축하고 차선 폐쇄로 인한 변화를 정량적으로 분석한다. 이를 위하여 360° 카메라, 차량 전방 카메라, 라인 스캔 카메라를 통하여 도로 영상을 수집하고, Mask R-CNN과 RF DETR+SAM 알고리즘을 활용하여 도로포장과 입체 구조물의 손상을 탐지하였다. 또한, 교통 시뮬레이션 프로그램 SUMO를 통해 국내 도로 구간을 재현하고 차량 에너지 분석 모듈 FASTSim을 연계하여 차선 폐쇄에 따른 교통 및 에너지 효율 변화를 비교하였다. AI 탐지 결과 RF DETR+SAM 시스템은 정확도 81%, 정밀도 87%, 재현율 61%, F1-score 0.72를 달성해 Mask R-CNN 대비 우수한 성능을 기록했으며, 도로포장뿐만 아니라 입체 구조물에 대한 안정적 탐지 가능성을 확인하였다. 시뮬레이션 결과 차선 폐쇄는 주행 속도 약 25% 감소, 연료 소모 약 18% 증가, CO2 배출량이 약 22% 증가한 것으로 나타났다. 본 연구는 AI 기반의 손상 탐지가 차량흐름을 유지하며 수행될 수 있음을 실증하고, 유지관리 시 교통, 에너지, 환경 영향을 통합적으로 고려할 수 있는 정량적 근거를 제시한다.

국내 노후 교량의 증가에 따라 유지관리 비용과 사회적 피해가 지속적으로 확대되고 있으며, 특히 포트홀 발생으로 인한 피해 보상액 또한 최근 증가하는 추세를 보이고 있다. 교량 포장 구조에서 포트홀은 아스팔트 포장과 콘크리트 바닥판 사이 계면의 박리로부터 구조적으로 시작된다. 차량 제동 및 가속에 따른 수평 하중, 수분 침투, 층간 차등 팽창 등은 계면에 인 장응력을 유발하여 결합 상태를 약화시키며, 이는 표면 균열로 진전되어 최종적으로 포트홀로 이어진다. 따라서 계면 박리는 포트홀 발생의 구조적 전조증상으로 볼 수 있다. 하지만 기존의 육안 점검은 표면 손상 중심의 평가에 국한되어 계면 박리 와 같은 내부 구조 상태를 직접적으로 파악하는 데 한계가 있다. 최근에는 구조물 내부 상태를 평가하기 위해 다양한 NDT 기법의 활용이 증가하고 있으나, 탄성파 기반의 IE(Impact-Echo) 및 UT(Ultrasonic Testing) 기법은 아스팔트와 같은 다공성 재료 내부에서 신호 감쇠가 발생하여 적용에 제약이 있다. 반면, 전자기파를 활용하는 GPR(Ground Penetrating Radar)은 포 장 내부 및 계면 상태 평가에 적합하나, 신호 해석 과정에서 전문가의 경험에 의존하는 주관적 한계가 존재한다. 이에 본 연구에서는 GPR 데이터를 기반으로 계면 박리 유무를 자동으로 분류하고, 이를 통해 포트홀 발생 위험 지점을 예측하는 딥 러닝 기반 프레임워크를 제안하였다. ResNet-50을 백본으로 하는 2-stage 전이학습 기법을 적용하였으며, 1단계에서는 3,708 개의 시험체 데이터를 활용하여 기초 분류 모델을 구축하고, 2단계에서는 28,890개의 실교량 데이터를 추가 학습하여 현장 조건에 대한 일반화 성능을 향상시켰다. 그 결과, 제안된 모델은 전체 정확도 85.2%와 weighted F1-score 0.8493의 성능을 나 타내었다. 본 연구에서 제안한 방법은 포트홀 발생 이전 단계에서 내부 계면 박리를 탐지할 수 있는 기술적 기반을 제시하 였으며, 이를 통해 선제적 유지관리 전략 수립과 교통 안전성 향상, 유지관리 비용 및 피해 보상액 감소에 기여할 수 있을 것으로 판단된다.

A dual-analyte electrochemical platform was developed using RuS2-Fe nanodots and a multi-walled carbon nanotube (MWCNT) incorporated RuS2-Fe composite (RuS2/MWCNT-Fe) composites for the sensitive detection of xylazine hydrochloride (XLZ) and erythrosine B (ERY). Both the RuS2-Fe nanodots and RuS2/MWCNT-Fe composites were synthesized via hydrothermal method then used to develop sensors via drop casting on glassy carbon electrodes (GCE). The RuS2-Fe nanodots and RuS2/MWCNT-Fe composites greatly improved the redox capacity of the interfacial region and electron transfer to the surface of the electrodes. Theoretical density functional calculations also validated experimental evidence of charge redistribution within the iron centres of the complex, narrowing of the band gap, and preferential adsorption of both XLZ and ERY. In particular, RuS2-Fe/GCE exhibited unprecedented electrodes within the context of the XLZ analyte, achieving 0.249 nM LOD and a linear range of 0.005–2500 μM. In contrasting work, RuS2/MWCNT-Fe composites electrode obtained 36 nM LOD and ranged 0.05–100 μM towards ERY. Careful analysis of electrochemical impedance and control studies utilizing pristine RuS2 with variable Fe concentrations, alongside extensive durability analysis, elucidated the significant influence of trace Fe concentrations on catalytic activity enhancements. In the context of recent reports on MXene, CNT, and oxide hybrids, the RuS2/MWCNT-Fe system still exhibited ample confirmations on charge transfer resistance and sensitivity. Proposed oxidation mechanisms illustrate the influence of iron on interfacial electron-proton coupling. The versatility of RuS2-Fe nanodots set as a carbon-based electrocatalyst has now been expanded to the dual detection of veterinary sedatives and food colorants. Such a development can be translated as a new stride toward the development of portable food safety, pharmaceutical quality control, and clinical diagnostics devices.

상수관망의 누수는 수자원 손실 및 시설물 피해의 주요 원인으로, 효과적인 탐지를 위해 다양한 기술이 개발되고 있다. 본 연구는 실제 상수도 누수 사례를 대상으로 청음 데이터를 수집하고, 주파수 및 청각 기반 음향 특징을 추출하여 비지도 학습 기반의 이상 감지 모델을 적용함으로써 누수음을 탐지하는 기법을 제안한다. 청음 신호에 대해 푸리에 변환과 멜 주파수 켑스트럼 계수(MFCC)를 적용하여 총 86개의 음향 특징을 구성하였으며, 랜덤 포레스트를 통해 주요 변수 6개를 선정하였다. 거리 기반 군집 분석을 통해 정상 소음 분포를 구성하고, Isolation Forest 및 Autoencoder 알고리즘을 활용하여 이상 음향을 판별하였다. 비지도 모델에 의해 탐지된 이상 지점이 실제 현장 판단과 부합함을 확인하였다. 정량적 기준 기반의 이상 탐지 결과가 실제 누수 지점과 일관성을 보였다. 본 연구는 실무 적용 가능한 비지도 이상 탐지 접근법을 제시함으로써, 기존 라벨 의존 탐지 방식의 한계를 보완할 수 있음을 시사한다.

배수 관망에서 가지관은 주 송수관에서 물을 주민에게 공급하는 핵심 요소이나, 일부는 불법적으로 사용되거나 정보가 불확실하여 관망 운영 효율을 저하시킨다. 본 연구는 단순 관망을 대상으로 천이류 해석과 실험을 통해 다중 가지관 탐지 기법을 검증하였다. 연구목적은 천이류 기반 가지관 탐색법의 적용 가능성을 평가하고, 관로 매개변수의 민감성을 분석하는 것이다. 개발된 방법은 단순 관로에서 전통적인 특성선 방법에 볼밸브의 비선형 거동 분석을 연계해 해석했다. 해석과 실험은 동일한 관망에서 두 개의 가지관을 대상으로 수행되었으며, 실험 내재 불확실성을 고려하였다. 두 분석 모두에서 구별 가능한 압력 신호가 확인되어 제안된 방법의 가지관 탐지 가능성을 입증하였다. 또한 결과는 불균일한 파속도와 일관되지 않은 천이류 유입 조건이 탐지 성능에 큰 영향을 미치는 것으로 나타났다.

Water contamination caused by heavy metal pollutants from industrial activities remains a pressing environmental concern. This study reports the development of a novel carbon paste electrode (CPE) modified with ethylenediaminetetraacetic acid (EDTA), polyvinyl alcohol (PVA), and multi-walled carbon nanotubes (MWCNTs) using a mechanochemical method for the electrochemical detection of Cu(II) ions. The modified electrode was thoroughly characterized to evaluate its functional groups, morphology, crystallinity, elemental composition, and electrochemical properties. Electrochemical measurements were performed using cyclic voltammetry (CV) and square-wave anodic stripping voltammetry (SWASV) under optimized conditions in 0.1 M NH₄Cl at pH 5. The EDTA/PVA/MWCNT-CPE exhibited a low detection limit (0.0457 μM), a wide linear range (0.1–2.7 μM), and excellent reproducibility (RSD = 0.51%), repeatability (RSD = 0.43%), and stability (95% retention after six days). Selectivity tests demonstrated high recovery for Cu(II) (99.7%) and Hg(II) (99.89%) with minimal interference. This simple, cost-effective sensor offers high sensitivity and selectivity, making it a promising candidate for Cu(II) detection in environmental monitoring applications.

With advancements in high-resolution scanners and high-performance computers, the use of whole slide imaging (WSI) in digital pathology has increased. WSI scans glass slides and stores them in digital format, making them immune to damage or discoloration, and enabling remote pathology review and peer review. Additionally, with the development of artificial intelligence, research using deep learning models in pathology has become more widespread. In this study, the You Only Look Once (YOLO)v8 model was used to train artificial intelligence to detect apoptotic bodies commonly observed in rodent livers. A total of 1,558 rat liver images containing apoptotic bodies were collected and followed by labeling and data augmentation using flipping and rotation techniques to expand the dataset to 3,738 images. The dataset was then divided into training, validation, and test sets to develop and evaluate a model for object recognition. The training was conducted with an epoch set to 300. The YOLOv8 model detected apoptotic bodies with a mean average precision at 50% value of 0.882. Although the model’s accuracy for detecting individual apoptotic bodies may not seem extremely high, it is important to note that the size of apoptotic bodies is very small compared to hepatocytes, making them harder to detect. However, the model’s overall performance is expected to improves significantly with a larger dataset. The YOLOv8 model successfully detected apoptotic bodies with high accuracy. This can help reduce the workload of toxicologic pathologists and decrease the time and cost involved in pathology review. Furthermore, with an increased dataset, even higher accuracy can be expected in the future.

Defect detection in manufacturing processes is a critical requirement for ensuring product reliability and maintaining production stability. As smart manufacturing environments continue to advance, the need for precise and robust vision-based inspection methods has become increasingly significant. This study proposes a hybrid defect analysis framework that integrates YOLOv5-based defect candidate detection with an Attention U-Net–based segmentation module. Experiments conducted on chromate-coated industrial images demonstrate that the proposed framework achieves an accuracy of 0.97, precision of 0.91, recall of 0.89, F1-score of 0.93, and IoU of 0.88, exhibiting stable performance even for small defects and irregular boundaries. The combination of region- of-interest extraction and attention-enhanced pixel-level segmentation improves both computational efficiency and boundary reconstruction quality. The findings extend the applicability of attention-based segmentation to industrial defect inspection and provide practical insights for deploying deep learning–based quality monitoring systems in automated manufacturing environments.