• Tribology and Lubricants
• /
• v.38 no.6
• /
• pp.281-286
• /
• 2022

In this study, the effects of Elasto-hydrodynamic lubrication pressure on the critical shoulder height of raceway in an angular contact ball bearing were investigated. Both 3D contact analyses using an influence function and the EHL analysis were conducted for the contact geometry between the ball and raceways. The pressure distributions by 3D contact analysis and EHL analysis for an example bearing were compared. The effect of ellipse truncation on the minimum film thickness also investigated from EHL analysis. The critical shoulder height in the dry contact and the EHL state were compared for various applied loads. It is shown that when the ellipse truncation occurs, the pressure spike for the EHL conjunction is higher than that for the dry contact, and its location moves more inward of the contact center. The steep pressure gradients would increase the flow rate, so in order to maintain flow continuity a significant reduction in film thickness and an abrupt rise in pressure occurs in the edge of shoulder. Significant reduction of the minimum film thickness occurs near the edge of shoulder. The critical shoulder heights in the EHL state are calculated as higher values compared with in the dry contact. This results shows that the determination of critical shoulder height by the EHL analysis is more proper.

• International Journal of Internet, Broadcasting and Communication
• /
• v.14 no.1
• /
• pp.136-141
• /
• 2022

A very complex task in deep learning such as image classification must be solved with the help of neural networks and activation functions. The backpropagation algorithm advances backward from the output layer towards the input layer, the gradients often get smaller and smaller and approach zero which eventually leaves the weights of the initial or lower layers nearly unchanged, as a result, the gradient descent never converges to the optimum. We propose a two-factor non-saturating activation functions known as Bea-Mish for machine learning applications in deep neural networks. Our method uses two factors, beta (𝛽) and alpha (𝛼), to normalize the area below the boundary in the Mish activation function and we regard these elements as Bea. Bea-Mish provide a clear understanding of the behaviors and conditions governing this regularization term can lead to a more principled approach for constructing better performing activation functions. We evaluate Bea-Mish results against Mish and Swish activation functions in various models and data sets. Empirical results show that our approach (Bea-Mish) outperforms native Mish using SqueezeNet backbone with an average precision (AP50val) of 2.51% in CIFAR-10 and top-1accuracy in ResNet-50 on ImageNet-1k. shows an improvement of 1.20%.

• Journal of Korean Port Research
• /
• v.9 no.2
• /
• pp.25-38
• /
• 1995

The wind generated circulation model describes the phenomenon based on the following physical assumptions: a) As the horizontal dimension of the flow domain is several orders of magnitude larger than vertical dimension, nearly horizontal flow is realistic. b) The time taken for circulation to develop may effect on the flow domain of the earth's rotation, the contribution of the Coriolis force. c) A flow domain of large dimension results in quite large Reynolds number and the Reynolds stresses are approximated by the turbulent mean velocity gradient. d) The circulation is forced by the shear stresses on the water surface exercised by the wind. Modification made to the depth average approximation of the convective terms and the bed shear stress terms by adopting a certain distribution of current over the depth and laboratory measurements for the bed shear expression. Modification circulation patterns, energy evolution and surface profile gave the significant differences comparing with the classical model results. The modified model results in higher free surface gradients balancing both the free surface shear and the bed shear and consequently to higher surface profiles along the coast.

• Wind and Structures
• /
• v.34 no.3
• /
• pp.303-312
• /
• 2022

As wind turbine rotors increase, the overall loads and dynamic response become an important issue. This problem is augmented by the exposure of wind turbines to severe atmospheric events with unconventional flows such as tornadoes, which need specific designs not included in standards and codes at present. An experimental study was conducted to analyze the loads induced by a tornado-like vortex (TLV) on horizontal-axis wind turbines (HAWT). A large-scale tornado simulation developed in The Wind Engineering, Energy and Environment (WindEEE) Dome at Western University in Canada, the so-called Mode B Tornado, was employed as the TLV flow acting on a rigid wind turbine model under two rotor operational conditions (idling and parked) for five radial distances. It was observed that the overall forces and moments depend on the location and orientation of the wind turbine system with respect to the tornado vortex centre, as TLV are three-dimensional flows with velocity gradients in the radial, vertical, and tangential direction. The mean bending moment at the tower base was the most important in terms of magnitude and variation in relation to the position of the HAWT with respect to the core radius of the tornado, and it was highly dependent on the rotor Tip Speed Ratio (TSR).

• Geophysics and Geophysical Exploration
• /
• v.25 no.4
• /
• pp.242-251
• /
• 2022

Automatic differentiation automatically calculates the derivatives of a function using the chain rule once the forward operation of a function is defined. Given the recent development of computing libraries that support automatic differentiation, many researchers have adopted automatic differentiation techniques to solve geophysical inverse problems. We analyzed the advantages, disadvantages, and performances of automatic differentiation techniques using the gradient calculations of seismic full waveform inversion objective functions. The gradients of objective functions can be expressed as multiplications of the derivatives of the model parameters, wavefields, and objective functions using the chain rule. Using numerical examples, we demonstrated the speed of analytic differentiation and the convenience of complex gradient calculations for automatic differentiation. We calculated derivatives of model parameters and objective functions using automatic differentiation and derivatives of wavefields using analytic differentiation.

• Geomechanics and Engineering
• /
• v.33 no.3
• /
• pp.317-326
• /
• 2023

Accurately measuring the hydraulic conductivity of dredged slurry (HCODS) is a difficult task and usually requires highly developed experimental techniques. To resolve such problem, this paper presents a novel laboratory method, where a double drainage sedimentation test (DDST) is proposed to generate a downward seepage after the end of primary consolidation (EOP). Based on the established stress equilibrium equations, it is figured out that the determination of local hydraulic gradients requires the effective stress distribution to be measured. Accordingly, an additional single drainage sedimentation test (SDST) with the same initial water content is performed in the novel laboratory method, which can be utilized to establish the relationship between effective stress and water content for investigated slurry. Thus, HCODS can be determined via a pair of SDST and DDST, with the water contents after the EOP measured. The corresponding calculation procedure is given in details. With a simply-designed settling column, the hydraulic conductivity tests were performed on three types of dredged slurry. The results demonstrated the effectiveness of the novel laboratory method in measuring HCODS.

• Journal of the Korean Society of Visualization
• /
• v.21 no.3
• /
• pp.75-84
• /
• 2023

In this study, we conducted analysis of bubble dynamics and flow of liquid phase change material(PCM) using shadowgraphy and particle image velocimetry(PIV). Characteristics of internal flow varied depending on locations of injection when solid PCM was liquefied from heated vertical wall. When bubbles rose immediately, they exhibited elliptical shape and zigzag trajectory. In contrast, when bubbles rose after merging at the bottom of solid PCM, with equivalent diameter for the inter-wall distance of 0.64 or greater, they showed a jellyfish shape and strong rocking behavior. It was observed by the PIV that the small ellipse bubbles made most strong flow inside the liquid PCM. Furthermore, the flow velocity was highest in the case of front injection, as the directions of temperature gradients and bubble-driven flow were aligned. The results underscore the significant influence of injection location on various characteristics, including bubble size, shape, rising path of bubbles, and internal flow.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.19 no.2
• /
• pp.91-99
• /
• 2024

Collision avoidance system is important to improve the robustness and functional safety of autonomous vehicles. This paper proposes an object-level distance estimation method to develop a collision avoidance system, and it is applied to golfcarts utilized in country club environments. To improve the detection accuracy, we continually trained an object detection model based on pseudo labels generated by a pre-trained detector. Moreover, we propose object-aware depth estimation (OADE) method which trains a depth model focusing on object regions. In the OADE algorithm, we generated dense depth information for object regions by utilizing detection results and sparse LiDAR points, and it is referred to as object-aware LiDAR projection (OALP). By using the OALP maps, a depth estimation model was trained by backpropagating more gradients of the loss on object regions. Experiments were conducted on our custom dataset, which was collected for the travel distance of 22 km on 54 holes in three country clubs under various weather conditions. The precision and recall rate were respectively improved from 70.5% and 49.1% to 95.3% and 92.1% after the continual learning with pseudo labels. Moreover, the OADE algorithm reduces the absolute relative error from 4.76% to 4.27% for estimating distances to obstacles.

• Korean Journal of Radiology
• /
• v.2 no.3
• /
• pp.121-131
• /
• 2001

Echocardiography and catheterization angiography suffer certain limitations in the evaluation of congenital heart diseases in adults, though these are overcome by MRI, in which a wide field-of view, unlimited multiplanar imaging capability and three-dimensional contrast-enhanced MR angiography techniques are used. In adults, recently introduced fast imaging techniques provide cardiac MR images of sufficient quality and with less artifacts. Ventricular volume, ejection fraction, and vascular flow measurements, including pressure gradients and pulmonary-to-systemic flow ratio, can be calculated or obtained using fast cine MRI, phase-contrast MR flow-velocity mapping, and semiautomatic analysis software. MRI is superior to echocardiography in diagnosing partial anomalous pulmonary venous connection, unroofed coronary sinus, anomalies of the pulmonary arteries, aorta and systemic veins, complex heart diseases, and postsurgical sequelae. Biventricular function is reliably evaluated with cine MRI after repair of tetralogy of Fallot, and Senning's and Mustard's operations. MRI has an important and growing role in the morphologic and functional assessment of congenital heart diseases in adolescents and adults.

• ETRI Journal
• /
• v.46 no.4
• /
• pp.648-659
• /
• 2024

Author verification is challenging because of the diversity in writing styles. We propose an enhanced handwriting verification method that combines handcrafted and automatically extracted features. The method uses a genetic algorithm to reduce the dimensionality of the feature set. We consider offline Bangla handwriting content and evaluate the proposed method using handcrafted features with a simple logistic regression, radial basis function network, and sequential minimal optimization as well as automatically extracted features using a convolutional neural network. The handcrafted features outperform the automatically extracted ones, achieving an average verification accuracy of 94.54% for 100 writers. The handcrafted features include Radon transform, histogram of oriented gradients, local phase quantization, and local binary patterns from interwriter and intrawriter content. The genetic algorithm reduces the feature dimensionality and selects salient features using a support vector machine. The top five experimental results are obtained from the optimal feature set selected using a consensus strategy. Comparisons with other methods and features confirm the satisfactory results.