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.

Heteroatom nitrogen-doped defect engineering is considered an effective strategy for enhancing the microwave absorption performance of carbon-based hybrid materials. The focus of present work is to study the electromagnetic absorption properties of as-prepared Co0.1Ni0.4Zn0.5Fe2O4/NrGO/MWCNT (CNZF/NrGO/MWCNT) nanocomposites that could be facilely modulated by changing the doping nitrogen contents to create the defect-induced polarisations which can be utilised as a promising candidates for microwave absorption materials (MAMs) for high-frequencies. Moreover, CNZF/ NrGO/MWCNT nanocomposites are designed using a facile one-pot solvothermal method by varying nitrogen-doped content and configuration. It was found that the optimisation between pyrrolic-N and graphitic-N, rather than total nitrogen content unlike earlier reports, plays a decisive role in regulating the electron magnetic properties. The micromorphological analysis reveals the presence of interfacial defects within the as-prepared samples, which are beneficial for electromagnetic attenuation. The optimally nitrogen-doped CNZF/NrGO/MWCNT composite (S2), prepared using ethylenediamine (EDA=2 ml), exhibits the balanced dielectric magnetic loss behaviour and enhanced interfacial polarisation, exhibited superior microwave absorption performance, achieving a minimum reflection loss (RLmin) of -56.39 dB at 13.05 GHz with a thickness of 1.5 mm and a maximum effective absorption bandwidth (EABmax) of 4.21 GHz in Ku band. Thus, this work provides a scalable and cost-effective method through defect engineering for designing an efficient MAMs for high-frequency 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.

Mn-based catalysts like hopcalite (Cu–Mn oxide) are widely studied for low-temperature CO oxidation, with efforts focused on enhancing their redox properties. Incorporating defect-free graphene as a support has shown promise in improving both structural and catalytic performance, making the development of scalable graphene-supported Cu–MnOx (Gr/Cu–MnOx) composites highly desirable. In this study, fluid flow control systems were effectively employed to produce exfoliated graphene sheets, which were subsequently utilized for synthesizing Gr/Cu–MnOx composite catalysts. The enhanced shear stress and mass transfer within the fluid flow system improved the textural properties of the composite catalysts, resulting in higher surface areas and pore volumes compared to those of the unmodified Cu–MnOx composite. The Gr/Cu–MnOx composite catalysts exhibited superior toluene removal performance, achieving a T90 value of 200 °C, surpassing the T90 value of 250 °C of the unmodified Cu–MnOx composite. Furthermore, the water resistance was assessed by evaluating the catalytic performance after exposure to 5 vol% water vapor. The presence of hydrophobic graphene in Gr/Cu–MnOx enhanced water resistance compared to that of unmodified Gr/Cu–MnOx.

상수관로의 노후화는 수질 안전성 저하와 수자원 손실, 유지보수 비용 증가 등의 문제를 야기하며, 이에 따라 지중 매설관의 상태를 신속하고 정확하게 진단할 수 있는 기술의 중요성이 커지고 있다. 특히 내시경 영상을 활용한 관로 점검은 가장 보편적인 방식으로 자리 잡았으나, 판독자의 숙련도에 따라 해석 편차가 발생하고, 대량 데이터의 신속한 처리에는 한계가 있다. 이러한 배경에서 본 연구는 관종⋅관경⋅용도 등 상수관 메타데이터를 모델에 통합하고, 관로 내 결함의 존재 여부와 유형, 크기를 동시에 예측할 수 있는 다중과제 학습(Multi-task Learning) 기반 인공지능 모델을 제안한다. 제안한 모델은 두 개의 예측 헤드를 통해 결함 판별과 정량적 분류를 병행하도록 설계되었으며, SHAP 기반 분석을 통해 모델의 판단 근거가 상수관로의 실제 결함 특성과 일치함을 확인하였다. 이러한 접근은 수작업 판독의 부담을 경감하고, 관로 상태 기록의 표준화 및 정량화를 통해 예방 중심의 유지관리 전략 수립을 효과적으로 지원할 수 있다.

Extensive soft tissue defects involving loss of skin, fat, and muscle often result from trauma or tumor resection. Current treatments, including autografts and flaps, are limited by donor-site morbidity and scarce tissue availability. Animal models, particularly in rodents, are essential for research but are limited by their primary healing mechanism—contraction via the panniculus carnosus—which does not accurately reflect human healing. Furthermore, standardized models for complex skin–muscle defects are lacking. Therefore, this study aims to create a clinically relevant composite soft tissue defect model in mice using a three-dimensional (3D) polylactic acid (PLA) chimney splint to inhibit contraction and better mimic human wound healing mechanisms (re-epithelialization and granulation tissue formation). A composite defect was created on the dorsum of 8-week-old BALB/c nude mice. The biocompatibility of the 3D-printed PLA chimney was assessed via MTT assay. In vivo, fixation methods—tissue adhesive (TA), simple interrupted sutures (SI), and purse-string suture (PS)—were compared. Wound healing was evaluated over 4 weeks via gross and histological analyses. PLA material showed excellent biocompatibility in vitro, with cell viability consistently above 85%, indicating noncytotoxicity. In vivo, the TA and SI groups showed severe inflammation, tissue necrosis, and splint detachment. In contrast, the PS group remained stable for 4 weeks with no complications. Histologically, the PS group effectively suppressed contraction. Re-epithelialization from the wound edge, well-organized granulation tissue with active angiogenesis, abundant fibroblasts, and collagen deposition, and spindle-shaped cells were clearly observed. In conclusion, this study establishes a reproducible and stable murine composite soft tissue defect model by combining a 3D-printed chimney splint with a PS technique. This model overcomes a key limitation of rodent wound models by controlling contraction, offering a robust preclinical platform to study composite tissue healing and evaluate next-generation regenerative medicine therapies.

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.

In the production sites of small and medium sized manufacturing enterprises, the increasing proportion of foreign workers has led to frequent difficulties in responding promptly to process defects and equipment setting errors during night and weekend shifts due to the absence of Korean supervisors. If such issues are not addressed in a timely manner, they can lead to large scale defects and reduced production efficiency. In this study, we developed an AI-based defect prediction and prevention system for the bearing machining process to overcome these on site management limitations. Real time machining data, equipment information, and quality inspection results were collected from the production lines of the target company, and the prediction accuracy of three models, RNN(Recurrent Neural Network), LSTM(Long Short-Term Memory), and GRU(Gated Recurrent Unit), was compared. As a result, the LSTM model demonstrated the best performance. The developed system visualizes real time defect prediction results in the form of a dashboard, enabling workers to immediately detect anomalies and adjust the process accordingly. Particularly in bearing machining processes where mass production occurs in short periods, the risk of lot level defects is high, while this system can contribute to improved production quality and efficiency by enabling early defect prediction and immediate response.

With the increasing number of aging buildings, the importance of structural safety inspections has grown significantly. Traditional methods for inspecting welding defects, such as visual inspection and magnetic testing, rely heavily on human expertise, making them time-consuming, costly, and subjective. To address these limitations, thermographic technology has been introduced as a non-contact alternative, significantly reducing both time and cost. Furthermore, by incorporating AI, an objective and automated evaluation of welding defects can be achieved. In this study, we propose an AI-based thermographic approach for detecting welding defects. To validate the applicability of this method, a Mock-up Test was conducted. Specifically, 12 types of welding specimens with 4 welding part were prepared, generating a dataset of 6,500 thermographic images. Among 7 regression algorithms tested, RF and EXT were selected due to their superior performance. By ensemble learning these two models, we developed a robust welding defect measurement algorithm. To further verify its effectiveness, we applied the developed algorithm to 2 real projects, evaluating its applicability using 450 thermographic images. The results of this study demonstrate the feasibility of AI and thermographic technology in welding defect detection, highlighting its potential to enhance the efficiency and reliability of structural safety inspections in aging infrastructures.

본 연구에서는 교목성 낙엽침엽수인 메타세쿼이아(Metasequoia glyptostroboides)가 가로수로 식재된 국내 8개 지역(삼척, 대전, 대구, 구미, 포항, 부산, 진안, 담양)의 10개 도로에서 총 280본을 대상으로 결함 및 관리 특성을 조사하였다. 육안 평가를 기반으로 2022년과 2023년 6~7월에 기본 현황, 결함, 관리 특성을 종합적으로 분석하였다. 결함도는 고사지, 줄기 상처, 병해충 등을 조사하여 정량화하였다. 메타세쿼이아의 평균 수고는 17.8m, 흉고직경 43.2cm, 근원직경 62.3cm, 수관폭 7.7m, 지하고 4.1m였으며, 흉고직경과 수고 간에는 전반적으로 양의 상관관계가 확인되었다. 그러나 흉고직경 기준 수고 예측 모델을 사용하였을 때 자연 집단보다 수고가 최대 11.9m 낮았다. 결함도는 평균 2.21점으로, 근계 결함(96.79%), 해충 피해(60.00%), 고사지(46.79%)가 가장 빈번했다. 보호틀 폭은 대부분 1~2m였으나 일부 구간은 1m 내외로 근계 손상이 발생하였고, 전선 비지중화 구간에서는 가지치기로 인해 수형이 고착되는 경향을 보였다. 교목성 가로수로서 전국적으로 조성된 메타세쿼이아 가로수의 지속 가능한 관리를 위한 종합적인 방안을 강구하는 것이 필요하다.

재건축아파트 하자담보책임 소송은 원고와 피고가 실질적으로 동일한 구성원 집단에 속할 수 있는 특수한 구조적 특성으로 인해 심각한 이해상충 문제를 내포하 고 있다. 본 연구는 이러한 구조적 모순이 소송의 공정성과 효율성에 미치는 영향을 분석하였다. 또한, 절차적 공정성 확보를 위한 개선방안을 모색하였다. 특히 조합원이자 구분소유자인 당사자들의 이중적 지위로 인해 전통적 대립당사자 주의 원칙이 훼손될 수 있다. 재건축아파트의 공용부분 재산 관리를 위한 총회결의 요건과 신속한 권리구제 필요성 간의 긴장관계가 소송 진행을 지연시킬 수도 있다. 이를 해결하기 위해 특별대리인 제도의 예방적 활용, 중립적 판단기구 도입, 외부 중립기관 개입 확대 등의 절차적 공정성 확보방안을 제시하였다. 또한 미국 HOA 제도와의 비교분석을 통해 입주자대표회의의 법적 지위 명확화, 포괄적 소송권한 부여 등의 입법적 개선방안을 도출하였다. 본 연구는 재건축하자 소송 제도의 구조적 문제 해결과 구분소유자 권익보호 강화에 기여할 것으로 기대된다.