• Journal of Ocean Engineering and Technology
• /
• v.34 no.5
• /
• pp.294-303
• /
• 2020

This study discusses data collection, calculation of wind and wave-induced resistance, and speed-power analysis of an 8,600 TEU container ship. Data acquisition system of the ship operator was improved to obtain the data necessary for the analysis, which was accomplished using SPA (Ship Performance Analysis, Park et al., 2019) in conformation with ISO15016:2015. From a previous operation profile of the container, the standard operating conditions of mean draft were 12.5 m and 13.6 m, which were defined with the mean stowage configuration of each condition. Model tests, including the load-variation test, were conducted to validate new ship performance and for the speed-power analysis. The major part of the added resistance of container ship is due to the wind. To check the reliability of wind-resistance calculation results, the resistance coefficients, added resistance, and speed-power analysis results using the Fujiwara regression formula (ISO15016:2015) and Computational fluid dynamics (Ryu et al., 2016; Jeon et al., 2017) analysis were compared. Wind speed and direction measured using an anemometer were used for wind-resistance calculation and the wave resistance was calculated using the wave-height and direction-data from weather information. Also, measured water temperature was used to calculate the increase in resistance owing to the deviation in water density. As a result, the SPA analysis using measured data and weather information was proved to be valid and able to identify the ship's resistance propulsion performance. Even with little difference in the air-resistance coefficient value, both methods provide sufficient accuracy for speed-power analysis. The differences were unnoticeable when the speed-power analysis results using each method were compared. Also, speed-power analysis results of the 8,600 TEU container ship in two draft conditions show acceptable trends when compared with the model test results and are also able to show power increase owing to hull fouling and aging. Thus, results of speed-power analysis of the existing 8,600 TEU container ship using the SPA program appropriately exhibit the characteristics of speed-power performance in deal conditions.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2011.05a
• /
• pp.14-14
• /
• 2011

Groundwater in the Waikatoregion is a valuable resource for agriculture, water supply, forestry and industries. The 434,000 ha study area comprises the upper Waikato River catchment from the outflow of Lake Taupo (New Zealand's largest lake) through to Lake Karapiro (a man-made hydro lake with high recreational value) (Figure 1). Water quality in the area is naturally high. However, there are indications that this quality is deteriorating as a result of land use intensification and deforestation. Compounding this concern for decision makers is the lag time between land use changes and the realisation of effects on groundwater and surface water quality. It is expected that the effects of land use changes have not yet fully manifested, and additional intensification may take decadesto fully develop, further compounding the deterioration. Consequently, Environment Waikato (EW) have proposed a programme of work to develop a groundwater model to assist managing water quality and appropriate policy development within the catchment. One of the most important and critical decisions of any modelling exercise is the choice of the modelling platform to be used. It must not inhibit future decision making and scenario exploration and needs to allow as accurate representation of reality as feasible. With this in mind, EW requested that two modelling platforms, MODFLOW/MT3DMS and FEFLOW, be assessed for their ability to deliver the long-term modelling objectives for this project. The two platforms were compared alongside various selection criteria including complexity of model set-up and development, computational burden, ease and accuracy of representing surface water-groundwater interactions, precision in predictive scenarios and ease with which the model input and output files could be interrogated. This latter criteria is essential for the thorough assessment of predictive uncertainty with third-party software, such as PEST. This paper will focus on the attributes of each modelling platform and the comparison of the two approaches against the key criteria in the selection process. Primarily due to the ease of handling and developing input files and interrogating output files, MODFLOW/MT3DMS was selected as the preferred platform. Other advantages and disadvantages of the two modelling platforms were somewhat balanced. A preliminary regional groundwater numerical model of the study area was subsequently constructed. The model simulates steady state groundwater and surface water flows using MODFLOW and transient contaminant transport with MT3DMS, focussing on nitrate nitrogen (as a conservative solute). Geological information for this project was provided by GNS Science. Professional peer review was completed by Dr. Vince Bidwell (of Lincoln Environmental).

• The Journal of the Acoustical Society of Korea
• /
• v.38 no.5
• /
• pp.485-493
• /
• 2019

The parametric array phenomenon refers to the generation of a high directivity low frequency wave from a small size radiation plate using the nonlinearity of the medium. In order to improve the usability of parametric array, the beam steering method of low frequency wave is researched, and the beam steering phenomenon is predicted easily using the PD (product directivity) model. However, the PD model can only be applied to Gaussian sources under quasi-linear conditions. Also, the prediction accuracy of low frequency wave beam width is poor. In this paper, a method for predicting the beam steering characteristics of a parametric array that can overcome the limitation of the PD model is investigated. For this purpose, the numerical analysis algorithm of the KZK (Khokhlov-Zabolotskaya-Kuzentsov) equation widely used for parametric array phenomenon prediction is improved. Thus, the beam steering characteristics are calculated by applying the electrical beam steering condition and comparing experimental results. As a result, the numerical analysis using the modified KZK equation algorithm in this study confirms that the beam steering phenomenon can be predicted even in a parametric array source that does not correspond to the quasi-linear condition.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.20 no.10
• /
• pp.125-132
• /
• 2019

Because the water exposed to shock waves caused by an underwater explosion cannot withstand the appreciable tension induced by the change in both pressure and velocity, the surrounding water is cavitated. This cavitating water changes the transferring circumstance of the shock loading. Three phenomena contribute to hull-plate damage; initial shock loading and its interaction with the hull plate, local cavitation, and local cavitation closure then shock reloading. Because the main concern of this paper is local cavitation due to a near-field underwater explosion, the water surface and the waves reflected from the sea bottom were not considered. A set of governing equations for the structure and the fluid were derived. A simple one-dimensional infinite plate problem was considered to verify this uncoupled solution approach compared with the analytic solution, which is well known in this area of interest. The uncoupled solution approach herein would be useful for obtaining a relatively high level of accuracy despite its simplicity and high computational efficiency compared to the conventional coupled method. This paper will help improve the understanding of fluid-structure interaction phenomena and provide a schematic explanation of the practical problem.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.4
• /
• pp.29-35
• /
• 2020

In this paper, we propose a new structured entity recognition DeNERT model. Recently, the field of natural language processing has been actively researched using pre-trained language representation models with a large amount of corpus. In particular, the named entity recognition, which is one of the fields of natural language processing, uses a supervised learning method, which requires a large amount of training dataset and computation. Reinforcement learning is a method that learns through trial and error experience without initial data and is closer to the process of human learning than other machine learning methodologies and is not much applied to the field of natural language processing yet. It is often used in simulation environments such as Atari games and AlphaGo. BERT is a general-purpose language model developed by Google that is pre-trained on large corpus and computational quantities. Recently, it is a language model that shows high performance in the field of natural language processing research and shows high accuracy in many downstream tasks of natural language processing. In this paper, we propose a new named entity recognition DeNERT model using two deep learning models, DQN and BERT. The proposed model is trained by creating a learning environment of reinforcement learning model based on language expression which is the advantage of the general language model. The DeNERT model trained in this way is a faster inference time and higher performance model with a small amount of training dataset. Also, we validate the performance of our model's named entity recognition performance through experiments.

• Journal of Advanced Navigation Technology
• /
• v.25 no.5
• /
• pp.426-432
• /
• 2021

Continuous wave (CW) Doppler radar has the advantage of being able to solve the privacy problem unlike camera and obtains signals in a non-contact manner. Therefore, this paper proposes a human activity recognition (HAR) system using CW Doppler radar, and presents the hardware design and implementation results for acceleration. CW Doppler radar measures signals for continuous operation of human. In order to obtain a single motion spectrogram from continuous signals, an algorithm for counting the number of movements is proposed. In addition, in order to minimize the computational complexity and memory usage, binarized neural network (BNN) was used to classify human motions, and the accuracy of 94% was shown. To accelerate the complex operations of BNN, the FPGA-based BNN accelerator was designed and implemented. The proposed HAR system was implemented using 7,673 logics, 12,105 registers, 10,211 combinational ALUTs, and 18.7 Kb of block memory. As a result of performance evaluation, the operation speed was improved by 99.97% compared to the software implementation.

• KIPS Transactions on Computer and Communication Systems
• /
• v.10 no.7
• /
• pp.191-198
• /
• 2021

Since the size of training dataset become large and the model is getting deeper to achieve high accuracy in deep learning, the deep neural network training requires a lot of computation and it takes too much time with a single node. Therefore, distributed deep learning is proposed to reduce the training time by distributing computation across multiple nodes. In this study, we propose hybrid allreduce strategy that considers the characteristics of each layer and communication and computational overlapping technique for synchronization of distributed deep learning. Since the convolution layer has fewer parameters than the fully-connected layer as well as it is located at the upper, only short overlapping time is allowed. Thus, butterfly allreduce is used to synchronize the convolution layer. On the other hand, fully-connecter layer is synchronized using ring all-reduce. The empirical experiment results on PyTorch with our proposed scheme shows that the proposed method reduced the training time by up to 33% compared to the baseline PyTorch.

• Journal of the Korean Geotechnical Society
• /
• v.35 no.4
• /
• pp.27-35
• /
• 2019

This study confirmed the applicability to the field of geotechnical engineering for relatively complicated space and many target design variables in back analysis. The Sharan's equation and the Blum's method were used for the tunnel field and the retaining wall as a model for the multi-variate problem of geotechnical engineering. Optimization methods are generally divided into a deterministic method and a stochastic method. In this study, Simulated Annealing Method (SA) was selected as a deterministic method and Differential Evolution Algorithm (DEA) and Particle Swarm Optimization Method (PSO) were selected as stochastic methods. The three selected optimization methods were compared by applying a multi-variate model. The problem of deterministic method has been confirmed in the multi-variate back analysis of geotechnical engineering, and the superiority of DEA can be confirmed. DEA showed an average error rate of 3.12% for Sharan's solution and 2.23% for Blum's problem. The iteration number of DEA was confirmed to be smaller than the other two optimization methods. SA was confirmed to be 117.39~167.13 times higher than DEA and PSO was confirmed to be 2.43~6.91 times higher than DEA. Applying a DEA to the multi-variate back analysis of geotechnical problems can be expected to improve computational speed and accuracy.

• The Journal of Korean Institute of Next Generation Computing
• /
• v.15 no.5
• /
• pp.64-74
• /
• 2019

Although a non-rigid registration has high demands in clinical practice, it has a high computational complexity and it is very difficult for ensuring the accuracy and robustness of registration. This study proposes a method of applying a non-rigid registration to 3D magnetic resonance images of brain in an unsupervised learning environment by using a deep-learning network. A feature vector between two images is produced through the network by receiving both images from two different patients as inputs and it transforms the target image to match the source image by creating a displacement vector field. The network is designed based on a U-Net shape so that feature vectors that consider all global and local differences between two images can be constructed when performing the registration. As a regularization term is added to a loss function, a transformation result similar to that of a real brain movement can be obtained after the application of trilinear interpolation. This method enables a non-rigid registration with a single-pass deformation by only receiving two arbitrary images as inputs through an unsupervised learning. Therefore, it can perform faster than other non-learning-based registration methods that require iterative optimization processes. Our experiment was performed with 3D magnetic resonance images of 50 human brains, and the measurement result of the dice similarity coefficient confirmed an approximately 16% similarity improvement by using our method after the registration. It also showed a similar performance compared with the non-learning-based method, with about 10,000 times speed increase. The proposed method can be used for non-rigid registration of various kinds of medical image data.

• Journal of Biomedical Engineering Research
• /
• v.42 no.4
• /
• pp.150-158
• /
• 2021

Intracranial hemorrhage (ICH) refers to acute bleeding inside the intracranial vault. Not only does this devastating disease record a very high mortality rate, but it can also cause serious chronic impairment of sensory, motor, and cognitive functions. Therefore, a prompt and professional diagnosis of the disease is highly critical. Noninvasive brain imaging data are essential for clinicians to efficiently diagnose the locus of brain lesion, volume of bleeding, and subsequent cortical damage, and to take clinical interventions. In particular, computed tomography (CT) images are used most often for the diagnosis of ICH. In order to diagnose ICH through CT images, not only medical specialists with a sufficient number of diagnosis experiences are required, but even when this condition is met, there are many cases where bleeding cannot be successfully detected due to factors such as low signal ratio and artifacts of the image itself. In addition, discrepancies between interpretations or even misinterpretations might exist causing critical clinical consequences. To resolve these clinical problems, we developed a diagnostic model predicting intracranial bleeding and its subtypes (intraparenchymal, intraventricular, subarachnoid, subdural, and epidural) by applying deep learning algorithms to CT images. We also constructed a visualization tool highlighting important regions in a CT image for predicting ICH. Specifically, 1) 27,758 CT brain images from RSNA were pre-processed to minimize the computational load. 2) Three different CNN-based models (ResNet, EfficientNet-B2, and EfficientNet-B7) were trained based on a training image data set. 3) Diagnosis performance of each of the three models was evaluated based on an independent test image data set: As a result of the model comparison, EfficientNet-B7's performance (classification accuracy = 91%) was a way greater than the other models. 4) Finally, based on the result of EfficientNet-B7, we visualized the lesions of internal bleeding using the Grad-CAM. Our research suggests that artificial intelligence-based diagnostic systems can help diagnose and treat brain diseases resolving various problems in clinical situations.