56016 research outputs found
Sort by
????????? ?????? ???????????? ?????? ????????? ?????? ?????? ????????? ?????? ??????
No full textDepartment of Industrial Engineeringclos
Computationally efficient control and local planning for the safe flight of UAVs
No full textDepartment of Mechanical EngineeringThis research aims to provide safety belts to unmanned aerial vehicles (UAVs). Numerous dangers exist on operating UAVs. Failures of flight control computer, sensors, communication, actuators, collision, unstable control system. This article concentrates on coping with actuator fault, collision, and performance of control system. A number of FTC and collision avoidance algorithms are developed, but most of them require heavy computational resources for optimization, mapping and others. In real application, the objectives of UAVs will be the commercial or military mission, which would need several functions such as object recognition, Simultaneous localization and mapping (SLAM), and etc. Furthermore, the limit of computational power is tighter compared to the ground vehicles because the payload is low. The performance of heavy algorithms can decrease and even utilizing mapping and complex path planning algorithms could be prohibited. Therefore, the objective of this thesis is providing safety with low computational efforts as a role of safety belt. In FTC, it is designed to be general and simple. Light weight collision avoidance algorithms are designed, ensuring the safety independent to the planning and mapping algorithms. As a result, UAVs can concentrate computing power more for other important missions. Also, it is expected to be available for small UAVs.clos
Demonstration of Electrochemical Device : Lab scale to pilot scale for Metal-CO2 system
No full textSchool of Energy and Chemical Engineering (Energy Engineering)clos
Development of high-strength nanoporous Cu via dealloying of intermetallic compound precursor
No full textDepartment of Materials Science and Engineeringclos
?????????????????? ?????? ????????? ?????????????????? ???????????????????????? ????????? ???????????? ?????? ??????
Get PDFDepartment of Nuclear EngineeringAfter the Fukushima nuclear accident in 2011, radioactive waste management measures and radioactive waste from nuclear power plants are attracting attention worldwide. In Korea, radioactive waste characterization and management measures are required as Kori Unit 1 is permanently shut down in 2017 and proceeds with an immediate decommissioning strategy. When decommissioning a nuclear facility, contaminated radioactive metals and concrete will account for most of the decommissioning radioactive waste. An unknown amount of radioactive waste including nuclides produced by unidentified impurities was produced during the decommissioning procedure required for the Kori Unit 1 nuclear power plant's (NPP) permanent shutdown. Therefore, when decommissioning a nuclear power plant, it is essential to predict the exposure dose of workers for radiation protection and to secure safety according to the radioactivity concentration and air pollution monitoring plan in the working area. To ensure such safety, analysis of the characteristics of decommissioning radioactive waste should be performed in a hot cell facility. Therefore, in this study, safety evaluation such as safety evaluation of hot cell facilities, workers' exposure dose evaluation, and radioactive material emission characteristics were performed in this process of handling highly uncertain waste. Additionally, the behavior of the airborne source term that occurs during the processing of radioactive waste when ventilation is considered in the hot cell facility was examined. The radioactive waste to be dealt with in the hot cell facility is composed of carbon steel and stainless steel as the reactor pressure vessel and reactor internal structure extracted from the reactor. More than 99% of the radioactivity inventory of the reactor pressure vessel and core barrel of the Pressurized Water Reactor (PWR) was composed of six nuclides60Co, 63Ni, 55Fe, 3H, 59Ni, and 14C. Additionally, the Monte Carlo N-Particle (MCNP) transport code was used to evaluate the workers' external exposure dose in the event of an accidental ventilation failure, a broken window, and normal operations at the hot cell facility during the processing of radioactive waste. The case where the highest dose value was obtained was evaluated when the reactor pressure vessel was processed with a circular saw. When the core beltline of reactor pressure vessel was cut using a circular saw, the case where the highest dose value was obtained was evaluated. The external exposure dose in the event of an unexpected ventilation failure was calculated at 2,950 times that of normal operation, and the external exposure dose because of a broken lead glass window was calculated as an additional 140 times. It was confirmed that the external exposure dose from the case of broken lead glass was 4.12E+05 times higher than the exposure dose from normal operation. Under the assumption that 1% of the period of the airborne source term will leak during normal operation, the internal exposure dose
of the worker was estimated to be 1.84 ?? 10-2 mSv/y, the highest in the same case. When worker exposed to contaminated air, the worker's exposure dose complied with the dose limit since it was confirmed that the worker was exposed by inhalation. In addition, the internal exposure dose evaluation was evaluated under the assumption that the airborne source term would flow into the working area in the event of an earthquake accident. At this time, the dose rate in the case of the highest dose value on the worker is 4.37 ?? 10-1 mSv/h. According to Article 14 of the Nuclear Safety and Security Commission Notification No. 2019-10, it was possible to infer the time to secure the safety of the worker. This enabled it to predict how long an irradiated worker could have access to their workspace prior to an accident. At this case, emergency working hours are allowed for a minimum of 1,145.47 hours, which should be sufficient to keep down the accident. Accordingly, a high efficiency particulate air filter(HEPA) with an efficiency of 99.97% was used to detect the emission concentration of the dominant nuclides (3H, 14C, 55Fe, 59Ni, 60Co, and 63Ni) of the airborne source term, which complies with the emission management standard of Nuclear Safety Commission Notice No. 2016-16. Based on these results, it is possible to evaluate the emission management in the airborne source term while processing the radioactive waste in a hot cell facility and performing radiation protection technology.ope
Flexible Adhesive Patch Based on Synergetic Crack Trapping Effects with Kirigami and Micropillar Hybrid Structures
No full textDepartment of Mechanical Engineeringclos
Highly reflective methyl acrylate/copolymer-TiO2 hybrid one-dimensional photonic crystal sensor for colorimetric detection of solvents in adulterated gasoline
Get PDFSchool of Energy and Chemical Engineering (Energy Engineering)Recently, the manufacture and distribution of adulterated gasoline mixed with pure commercial gasoline and adulterants has become a global problem. However, there is still a lack of a simple way to detect it immediately on-the-spot. In this study, we propose highly reflective one-dimensional photonic crystals (1D PC) as a simple, on-the-spot detection method for adulterated gasoline with the naked-eye. First, a photo cross-linked polymer???TiO2 hybrid material (Ti70) with a high refractive index and P(MA-co-BPA) with a low refractive index were synthesized and analyzed. After that, they were used to fabricate a total of 11 layers of highly reflective Ti70/P(MA-co-BPA) 1D PC. The fabricated Ti70/P(MA-co-BPA) 1D PC exhibited purple in pristine and no color change was observed when immersed in pure commercial gasoline. On the other hand, when immersed in methanol, xylene, toluene, and benzene, the colors of Ti70/P(MA-co-BPA) 1D PCs changed rapidly to strong blue, blue, yellow-green, and orange, respectively. In addition, Ti70/P(MA-co-BPA) 1D PC showed color change within a short time (~10 s) even in binary or ternary solvent mixtures(gasoline-toluene, gasoline-methanol, gasoline-toluene-methanol). Moreover, the excellent reusability of Ti70/P(MA-co-BPA) 1D PC was verified through 10 repetitions of swelling-deswelling tests. Through these results, it was confirmed that the novel Ti70/P(MA-co-BPA) 1D PC can be successfully applied to the colorimetric on-the-spot detection of adulterated gasoline.ope
Material-Correlated Film Deposition Engineering Toward Scalable Organic Solar Cells
Get PDFSchool of Energy and Chemical Engineering (Energy engineering)For a long period, organic solar cells (OSCs) has been attracted attention owing to the potentials in applicability with properties such as flexibility, light weight, and semitransparency. Also, the power conversion efficiency (PCE) of OSCs has been improved gradually by engineering the applied materials and fabrication process. Recently, the highest PCE over 19% was achieved in lab-scale by such steady efforts for optimizing material design and device fabrication. Owing to the almost saturated PCE, it is time to meet the requirements to be marketable: Reproducibility and sustainability. For reproducibility, there are two requirements. One is the change in fabrication method from spin-coating for lab-scale device to roll to roll process for bulk production. The other is that enlarged device area should be fabricated with comparable PCE to that of small area device. For sustainability, the OSCs have to be tolerable to various stress such as thermal stress, light, moisture, and aging by time. Therefore, my research was focused on the reproducibility of device fabrication with large area and device stability as well as photovoltaic performance.
The first study is aimed for the repeatable and large area device fabrication, so the spin coating process is replaced with blade coating process. The device performance from blade coating can be comparable to that from spin coating by controlling blade speed delicately. Also, solution process has an issue of evaporation driven flows, which cause mass transport and less uniformity. To achieve high quality film with sufficient homogeneity, the relationship between inner alkyl chain length of acceptor and the evaporation driven flow was studied with concept of surface energy and surface tension. As a results, the short inner alkyl chain acceptor was the favorable for large area device fabrication due to preferred direction of evaporation driven flow. The uniformity was confirmed by measuring thickness of active layer and surface topology at 9 spots on one substrate.
The second study is aimed for enhanced PCE and stability. It is important to maintain the constructed morphology under operation. Therefore, in this work, the acceptor with the ester-functionalized end group and the asymmetric structure was studied. Interestingly, the acceptors with the new end groups has the optimal condition without the thermal annealing treatment, exhibiting the higher PCE than that of Y6 based device, skipping post-treatment step. Moreover, the PCE and thermal stability was better than the representative acceptor, Y6, exhibiting improved device performances and less morphological changes under 85??? temperature on a hotplate. The third study is aimed for development in all polymer solar cells (APSCs). APSCs have shown their advantages in morphological stability under various stress such as light, heat, and mechanical force. NDI-based random copolymer acceptors were synthesized by substituting the accepting unit and investigated about the impact of symmetricity on morphological properties with AFM and TEM analysis. As a result, breaking symmetricity generally exhibited the higher FF with favored morphology.ope
Methods and Experiments for Understanding Two Interacting Hands
Get PDFGraduate School of Artificial IntelligenceThree dimensional hand pose estimation has reached a level of maturity, enabling real-world applications for single-hand cases. However, accurate estimation of the pose of two closely interacting hands still remains a challenge as in this case, one hand often occludes the other. We present a new algorithm that accurately estimates hand poses in such a challenging scenario. The crux of our algorithm lies in a framework that jointly trains the estimators of interacting hands, leveraging their inter-dependence. Further, we employ a GAN-type discriminator of interacting hand pose that helps avoid physically implausible configurations, e.g. intersecting fingers, and exploit the visibility of joints to improve intermediate 2D pose estimation. We incorporate them into a single model that learns to detect hands and estimate their pose based on a unified criterion of pose estimation accuracy. To our knowledge, this is the first attempt to build an end-to-end network that detects and estimates the pose of two closely interacting hands (as well as single hands). In the experiments with three datasets representing challenging realworld scenarios, our algorithm demonstrated significant and consistent performance improvements over state-of-the-arts.ope
Personal recommender system via convolutional autoencoder with conditioning augmentation : Recommender system and representation learning with convolutional autoencoder
Get PDFDepartment of Industrial EngineeringRecently, volume of various types of information, including reviews, images, and videos containing sound, also known as unstructured data, have been increased in an explosive manner . Even though, unstructured data is difficult to utilize without proper preprocessing, many applications adopted it as a source of information to extract value from it, such as recommender systems, natural language processing and computer visions . The recent prosperity of deep learning techniques has accelerated the progress in this field by making the preprocessing parts and feature extraction parts simple and easy . Moreover, the emergence of generative adversarial network has led to improved general performance of unsupervised learning models which makes many applications to make use of diverse forms of data
Along with this context, concerns of this article focused on (i) developing a recommender system based on modified autoencoder which is a typical deep learning technique applied to this research field that presents exceptional performance in the feature extraction process and (ii) applying data augmentation to this field which is frequently used to deal with data scarcity or limitation problem which is one of the main challenges of recommender systems [3].(iii) Lastly, the proposed model can be applied to both tasks, collaborative filtering and contents-based filtering proved to present compliant performance Modified convolutional autoencoder-based recommender system learns features of samples that represented by reviews of users or user-item rating matrices. The proposed model takes vanilla autoencoder as a base structure conbined with convolutional layer to extract feature that takes encoded vector as input which is represented by preprocessed reviews or ratings. Afterward, the conditioning augmentation process, which is an augmentation technique for embedded vectors, is applied that goes through a decoder that produces final predictions based on the encoded input vector from the previous encoder The contribution of the paper can be summarized as three points. (i)Conditioning augmentation which is data augmentation technique utilizing encoded vector to deal with
data scarcity problem that is mainly concerned in recommender system field. (ii)Proposed model can take both types of inputs which are contents-based encoded vector or ratingbased encoded vector. Contents-based vector can represent features of item such as review, quality, and various types of categorical feature to corresponding product. Rating-based vector indicates evaluation of consumer based on numeric value. (iii)Lastly, performance of proposed model is compliant compared to state of the arts of several bench mark open dataset.ope