• Smart Structures and Systems
• /
• v.29 no.1
• /
• pp.1-16
• /
• 2022

Despite recent breakthroughs in deep learning and computer vision fields, the pixel-wise identification of tiny objects in high-resolution images with complex disturbances remains challenging. This study proposes a modified U-net for tiny crack segmentation in real-world steel-box-girder bridges. The modified U-net adopts the common U-net framework and a novel Self-Attention-Self-Adaption (SASA) neuron as the fundamental computing element. The Self-Attention module applies softmax and gate operations to obtain the attention vector. It enables the neuron to focus on the most significant receptive fields when processing large-scale feature maps. The Self-Adaption module consists of a multiplayer perceptron subnet and achieves deeper feature extraction inside a single neuron. For data augmentation, a grid-based crack random elastic deformation (CRED) algorithm is designed to enrich the diversities and irregular shapes of distributed cracks. Grid-based uniform control nodes are first set on both input images and binary labels, random offsets are then employed on these control nodes, and bilinear interpolation is performed for the rest pixels. The proposed SASA neuron and CRED algorithm are simultaneously deployed to train the modified U-net. 200 raw images with a high resolution of 4928 × 3264 are collected, 160 for training and the rest 40 for the test. 512 × 512 patches are generated from the original images by a sliding window with an overlap of 256 as inputs. Results show that the average IoU between the recognized and ground-truth cracks reaches 0.409, which is 29.8% higher than the regular U-net. A five-fold cross-validation study is performed to verify that the proposed method is robust to different training and test images. Ablation experiments further demonstrate the effectiveness of the proposed SASA neuron and CRED algorithm. Promotions of the average IoU individually utilizing the SASA and CRED module add up to the final promotion of the full model, indicating that the SASA and CRED modules contribute to the different stages of model and data in the training process.

• Journal of Microbiology and Biotechnology
• /
• v.13 no.6
• /
• pp.993-997
• /
• 2003

Response surface methodology was applied to find the optimum conditions for the production of DXAMase (containing both dextranase and amylase activities) based on the cultivation variables (pH, temperature, and agitation rate). The experimental values from the model equation conceded with predicted values in which the predicted values for dextranase and amylase activities were 2.26 and 3.52 U/ml at pH 4, $28^{\circ}C$, 235 rpm, and the corresponding experimental values were 2.41 and 3.68 U/ml, respectively.

• Korean Journal of Mathematics
• /
• v.25 no.2
• /
• pp.279-301
• /
• 2017

An L(3, 2, 1)-labeling for the graph G = (V, E) is an assignment f of a label to each vertices of G such that ${\mid}f(u)-f({\upsilon}){\mid}{\geq}4-k$ when $dist(u,{\upsilon})=k{\leq}3$. The L(3, 2, 1)-labeling number, denoted by ${\lambda}_{3,2,1}(G)$, for G is the smallest number N such that there is an L(3, 2, 1)-labeling for G with span N. In this paper, we compute the L(3, 2, 1)-labeling number ${\lambda}_{3,2,1}(G)$ when G is a cylindrical grid, which is the cartesian product $P_m{\Box}C_n$ of the path and the cycle, when $m{\geq}4$ and $n{\geq}138$. Especially when n is a multiple of 4, or m = 4 and n is a multiple of 6, then we have ${\lambda}_{3,2,1}(G)=11$. Otherwise ${\lambda}_{3,2,1}(G)=12$.

• Restorative Dentistry and Endodontics
• /
• v.24 no.1
• /
• pp.67-75
• /
• 1999

The use of composite restorative materials is established due to continuing improvements in the materials and restorative techniques. Composite resins are widely used for the restoration of cervical lesions because of esthetics, good physical properties and working time. There are several types of cavity design for class V composite resin filling, but inappropriate cavity form may affect bonding failure, microleakage and fracture during mastication. Cavity preparations for composite materials should be as conservative as possible. The extent of the preparation is usually determined by the size, shape, and location of the defect. The design of the cavity preparation to receive a composite restoration may vary depending on several factors. In this study, 5 types of class V cavity were prepared on each maxillary central incisor. The types are; 1) V-shape, 2) round(U) shape, 3) box form, 4) box form with incisal bevel and 5) box form with incisal bevel and grooves for axial line angles. After restoration, in order to observe the concentration of stress at bonding surfaces of teeth and restorations, developing a 2-dimensional finite element model of labiopalatal section in tooth, surrounding bone, periodontal ligament and gingiva, based on the measurements by Wheeler, loading force from direction of 45 degrees from lingual side near the incisal edge was applied. This study analysed Von Mises stress with SuperSap finite element analysis program(Algor Interactive System, Inc.). The results were as follows : 1. Stress concentration was prevalent at tooth-resin bonding surface of cervical side on each model. 2. In model 2 without line angle, stress was distributed evenly. 3. Preparing bevel eliminated stress concentration much or less at line angle. 4. Model with round-shape distributed stress concentration more evenly than box-type model with sharp line angle, therefore decreased possibility of fracture. 5. Adding grooves to line angles had no effect of decreasing stress concentration to the area.

• Nuclear Engineering and Technology
• /
• v.27 no.3
• /
• pp.301-316
• /
• 1995

experimental and computational studies ore carried out to investigate the natural convection of the single phase flow in a tank with a heated horizontal plate facing downward. This is a simplified model for investigations of the influence of a core melt at the bottom of a reactor vessel on the thermal hydraulic behavior in a oater filled cavity surrounding the vessel. In this case the vessel is simulated by a hexahedron insulated box with a heated plate Horizontally mounted at the bottom of the box. The box with the heated plate is installed in a water filled hexahedron tank. Coolers are immersed in the U-type water volume between the box and the tank. Although the multicomponent flows exist more probably below the heated plate in reality, present study concentrates on the single phase flow in a first step prior to investigating the complicated multicomponent thermal hydraulic phenomena. In the present study, in order to get a better understanding for the natural convection characteristics below the heated plate, the velocity and temperature are measured by LDA(Laser Doppler Anemometry) and thermocouples, respectively. And How fields are visualized by taking pictures of the How region with suspended particles. The results show the occurrence of a very effective circulation of the fluid in the whole How area as the heater and coolers are put into operation. In the remote region below the heated plate the new is nearly stagnant, and a remarkable temperature stratification can be observed with very thin thermal boundary. Analytical predictions using the FLUTAN code show a reasonable matching of the measured velocity fields.

• Journal of Soil and Groundwater Environment
• /
• v.20 no.7
• /
• pp.43-52
• /
• 2015

The uranium removal capacity of radish sprouts (Raphanus sativus L.) in groundwater was calculated on the basis of the amount of uranium accumulated in the radish sprouts rather than the concentration in solution, of which process was very limited in previous studies. Continuous rhizofiltration clean-up system was designed to investigate the feasibility of radish sprouts, applying for uranium contaminated groundwater (U concentration: 110 μg/L) taken at Bugogdong, Busan. Six acrylic boxes (10 cm × 30 cm × 10 cm) were connected in a direct series for the continuous rhizofiltration system and 200 g of radish sprouts cultivars was placed in each box. The groundwater was flushed through the system for 48 hours at the constant rate of 5 mL/min. The rhizofiltration system was operated in the phytotron, of which conditions were at 25℃ temperature, 70% of relative humidity, 4,000 Lux illumination (16 hours/day) and 600 mg/L of CO₂ concentration. While 14.4 L of contaminated groundwater was treated, the uranium removal efficiency of the radish sprouts (1,200 g in wet weight) was 77.2% and their removal capacities ranged at 152.1 μg/g-239.7 μg/g (the average: 210.8 μg/g), suggesting that the radish sprouts belong to the group of hyper-accumulation species. After the experiment, the sum of U amounts accumulated in radish sprouts and remained in groundwater was 1,472.2 μg and the uranium recovery ratio of this rhizofiltration experiment was 92.9%. From the results, it was investigated that the radish sprouts can remove large amounts of uranium from contaminated groundwater in a short time (few days) because the fast growth rate and the high U accumulation adsorption capacity.

• Biotechnology and Bioprocess Engineering:BBE
• /
• v.6 no.4
• /
• pp.301-305
• /
• 2001

The co-expression of the arg U gene in a double-vector expression system of recombi-nant Escherichia coli BL22(DE3)[pET-IEN2a+pAC-argU] significantly enhanced the production level of reconminant human interferon -$\alpha$2a(rhIFN-$\alpha$2a) in high cell density cultures, compared to a recombinant E. coli culture containing only the single expression vector, pET-IEN2a. The dry cell mass concentration increased to almost 100 g/L, and more than 4 g/L of rhIFN-$\alpha$2a was accumu-lated in the culture broth. Evidently, the synthesis of rhIFN-$\alpha$2a was strongly dependent on the pre-induction growtih rate and more efficient at a higher specific growth rate. The additional sup-ply of tRN $A^{Arg(AGG/AGA)}$ enhanced the expression level of the rhIFN-$\alpha$2a gene in the early stage of the post-induction phase, yet thereafter the specific production rate of rhIFN-$\alpha$2a rapidly de-creased due to severe segregational instability of plasmid vector pET-IEN2a. It would appear that the plasmid instability with only occurred to pET-IEN2a in the double vector system, was re-lated to the effect of translational stress due to the over expression of rhIFN-$\alpha$2a.

• Structural Engineering and Mechanics
• /
• v.11 no.5
• /
• pp.519-534
• /
• 2001

The effect of the initial imperfections on the nonlinear behaviors and ultimate strength of the thin-walled members subjected to the axial loads, obtained by the finite element stability analysis, are examined. As the initial imperfections, the bucking mode shapes of the members are adopted. The buckling mode shapes of the thin-walled members are obtained by the transfer matrix method. In the finite element stability analysis, isoparametric degenerated shell element is used, and the geometrical and material nonlinearity are considered based on the Green Lagrange strain definition and the Prandtl-Reuss stress-strain relation following the von Mises yield criterion. The U-, box- and I-section members subjected to the axial loads are adopted for numerical examples, and the effects of the initial imperfections on the nonlinear behaviors and ultimate strength of the members are examined.

• Journal of the Korean Data and Information Science Society
• /
• v.28 no.5
• /
• pp.1191-1204
• /
• 2017

Weighted least squares (WLS) estimation is often easily used for the data with heteroscedastic errors because it is intuitive and computationally inexpensive. However, WLS estimator is less robust to a few outliers and sometimes it may be inefficient. In order to overcome robustness problems, Box-Cox transformation, Huber's M estimation, bisquare estimation, and Yohai's MM estimation have been proposed. Also, more efficient estimations than WLS have been suggested such as Bayesian methods (Cepeda and Achcar, 2009) and semiparametric methods (Kim and Ma, 2012) in heteroscedastic error models. Recently, Çelik (2015) proposed the weight methods applicable to the heteroscedasticity patterns including butterfly-distributed residuals and megaphone-shaped residuals. In this paper, we review heteroscedastic regression estimators related to robust or efficient estimation and describe their properties. Also, we analyze cost data of U.S. Electricity Producers in 1955 using the methods discussed in the paper.

• Communications for Statistical Applications and Methods
• /
• v.21 no.3
• /
• pp.213-224
• /
• 2014

In this paper, we study the problem of transforming to normality. We propose to estimate the transformation parameter by minimizing a weighted squared distance between the empirical characteristic function of transformed data and the characteristic function of the normal distribution. Our approach also allows for other symmetric target characteristic functions. Asymptotics are established for a random sample selected from an unknown distribution. The proofs show that the weight function $t^{-2}$ needs to be modified to have thinner tails. We also propose the method to compute the influence function for M-equation taking the form of U-statistics. The influence function calculations and a small Monte Carlo simulation show that our estimates are less sensitive to a few outliers than the maximum likelihood estimates.