• Geophysics and Geophysical Exploration
• /
• v.24 no.3
• /
• pp.89-97
• /
• 2021

Fault detection in seismic data is well suited to the application of machine learning algorithms. Accordingly, various machine learning techniques are being developed. In recent studies, machine learning models, which utilize synthetic data, are the particular focus when training with deep learning. The use of synthetic training data has many advantages; Securing massive data for training becomes easy and generating exact fault labels is possible with the help of synthetic training data. To interpret real data with the model trained by synthetic data, the synthetic data used for training should be geologically realistic. In this study, we introduce a method to generate realistic synthetic seismic data. Initially, reflectivity models are generated to include realistic fault structures, and then, a one-way wave equation is applied to efficiently generate seismic stack sections. Next, a migration algorithm is used to remove diffraction artifacts and random noise is added to mimic actual field data. A convolutional neural network model based on the U-Net structure is used to verify the generated synthetic data set. From the results of the experiment, we confirm that realistic synthetic data effectively creates a deep learning model that can be applied to field data.

• Korean Journal of Metals and Materials
• /
• v.49 no.11
• /
• pp.882-892
• /
• 2011

The influence of rhenium (Re) and ruthenium (Ru) addition on the solidification and solute redistribution behaviors in advanced experimental Ni-base superalloys has been investigated. A series of model alloys with different levels of Re and Ru were designed based on the composition of Ni-6Al-8Ta and were prepared by vacuum arc melting of pure metallic elements. In order to identify the influence of Re and Ru addition on the thermo-physical properties, differential scanning calorimetry analyses were carried out. The results showed that Re addition marginally increases the liquidus temperature of the alloy. However, the ${\gamma}^{\prime}$ solvus was significantly increased at a rate of $8.2^{\circ}C/wt.%$ by the addition of Re. Ru addition, on the other hand, displayed a much weaker effect on the thermo-physical properties or even no effect at all. The microsegregation behavior of solute elements was also quantitatively estimated by an electron probe microanalysis on a sample quenched during directional solidification of primary ${\gamma}$ with the planar solid/liquid interface. It was found that increasing the Re content gradually increases the microsegregation tendency of Re into the dendritic core and ${\gamma}^{\prime}$ forming elements, such as Al and Ta, into the interdendritic area. The strongest effect of Ru addition was found to be Re segregation. Increasing the Ru content up to 6 wt.% significantly alleviated the microsegregation of Re, which resulted in a decrease of Re accumulation in the dendritic core. The influence of Ru on the microstructural stability toward the topologically close-packed phase formation was discussed based on Scheil type calculations with experimentally determined microsegregation results.

• Geophysics and Geophysical Exploration
• /
• v.26 no.3
• /
• pp.114-125
• /
• 2023

The physics-informed neural network (PINN) has been proposed to overcome the limitations of various numerical methods used to solve partial differential equations (PDEs) and the drawbacks of purely data-driven machine learning. The PINN directly applies PDEs to the construction of the loss function, introducing physical constraints to machine learning training. This technique can also be applied to wave equation modeling. However, to solve the wave equation using the PINN, second-order differentiations with respect to input data must be performed during neural network training, and the resulting wavefields contain complex dynamical phenomena, requiring careful strategies. This tutorial elucidates the fundamental concepts of the PINN and discusses considerations for wave equation modeling using the PINN approach. These considerations include spatial coordinate normalization, the selection of activation functions, and strategies for incorporating physics loss. Our experimental results demonstrated that normalizing the spatial coordinates of the training data leads to a more accurate reflection of initial conditions in neural network training for wave equation modeling. Furthermore, the characteristics of various functions were compared to select an appropriate activation function for wavefield prediction using neural networks. These comparisons focused on their differentiation with respect to input data and their convergence properties. Finally, the results of two scenarios for incorporating physics loss into the loss function during neural network training were compared. Through numerical experiments, a curriculum-based learning strategy, applying physics loss after the initial training steps, was more effective than utilizing physics loss from the early training steps. In addition, the effectiveness of the PINN technique was confirmed by comparing these results with those of training without any use of physics loss.

• Geophysics and Geophysical Exploration
• /
• v.26 no.2
• /
• pp.37-51
• /
• 2023

The ground roll is the most common coherent noise in land seismic data and has an amplitude much larger than the reflection event we usually want to obtain. Therefore, ground roll suppression is a crucial step in seismic data processing. Several techniques, such as f-k filtering and curvelet transform, have been developed to suppress the ground roll. However, the existing methods still require improvements in suppression performance and efficiency. Various studies on the suppression of ground roll in seismic data have recently been conducted using deep learning methods developed for image processing. In this paper, we introduce three models (DnCNN (De-noiseCNN), pix2pix, and CycleGAN), based on convolutional neural network (CNN) or conditional generative adversarial network (cGAN), for ground roll suppression and explain them in detail through numerical examples. Common shot gathers from the same field were divided into training and test datasets to compare the algorithms. We trained the models using the training data and evaluated their performances using the test data. When training these models with field data, ground roll removed data are required; therefore, the ground roll is suppressed by f-k filtering and used as the ground-truth data. To evaluate the performance of the deep learning models and compare the training results, we utilized quantitative indicators such as the correlation coefficient and structural similarity index measure (SSIM) based on the similarity to the ground-truth data. The DnCNN model exhibited the best performance, and we confirmed that other models could also be applied to suppress the ground roll.

• Geophysics and Geophysical Exploration
• /
• v.25 no.4
• /
• pp.189-200
• /
• 2022

Ground subsidence on urban roads is a social issue that can lead to human and property damages. Therefore, it is crucial to detect underground cavities in advance and repair them. Underground cavity detection is mainly performed using ground penetrating radar (GPR) surveys. This process is time-consuming, as a massive amount of GPR data needs to be interpreted, and the results vary depending on the skills and subjectivity of experts. To address these problems, researchers have studied automation and quantification techniques for GPR data interpretation, and recent studies have focused on deep learning-based interpretation techniques. In this study, we described a hyperbolic event detection process based on deep learning for GPR data interpretation. To demonstrate this process, we implemented a series of algorithms introduced in the preexisting research step by step. First, a deep learning-based YOLOv3 object detection model was applied to automatically detect hyperbolic signals. Subsequently, only hyperbolic signals were extracted using the column-connection clustering (C3) algorithm. Finally, the horizontal locations of the underground cavities were determined using regression analysis. The hyperbolic event detection using the YOLOv3 object detection technique achieved 84% precision and a recall score of 92% based on AP50. The predicted horizontal locations of the four underground cavities were approximately 0.12 ~ 0.36 m away from their actual locations. Thus, we confirmed that the existing deep learning-based interpretation technique is reliable with regard to detecting the hyperbolic patterns indicating underground cavities.

• Geophysics and Geophysical Exploration
• /
• v.25 no.4
• /
• pp.227-241
• /
• 2022

Full waveform inversion (FWI) in the field of seismic data processing is an inversion technique that is used to estimate the velocity model of the subsurface for oil and gas exploration. Recently, deep learning (DL) technology has been increasingly used for seismic data processing, and its combination with FWI has attracted remarkable research efforts. For example, DL-based data processing techniques have been utilized for preprocessing input data for FWI, enabling the direct implementation of FWI through DL technology. DL-based FWI can be divided into the following methods: pure data-based, physics-based neural network, encoder-decoder, reparameterized FWI, and physics-informed neural network. In this review, we describe the theory and characteristics of the methods by systematizing them in the order of advancements. In the early days of DL-based FWI, the DL model predicted the velocity model by preparing a large training data set to adopt faithfully the basic principles of data science and apply a pure data-based prediction model. The current research trend is to supplement the shortcomings of the pure data-based approach using the loss function consisting of seismic data or physical information from the wave equation itself in deep neural networks. Based on these developments, DL-based FWI has evolved to not require a large amount of learning data, alleviating the cycle-skipping problem, which is an intrinsic limitation of FWI, and reducing computation times dramatically. The value of DL-based FWI is expected to increase continually in the processing of seismic data.

• Journal of the Microelectronics and Packaging Society
• /
• v.10 no.3
• /
• pp.19-27
• /
• 2003

Surface finishes of PCB laminates are important in the solder joint reliability of flip chip package because the types and thicknesses of intermetallic compound(IMC), and compositions and hardness of solders are affected by them. In this study, effects of surface finishes of PCB on the low cycle fatigue resistance of Sn-based lead-free solders; Sn-3.5Ag, Sn-3.5Ag-XCu(X=0.75, 1.5), Sn-3.5Ag-XBi(X=2.5, 7.5) and Sn-0.7Cu were investigated for the Cu and Au/Ni surface finish treatments. Displacement controlled room temperature lap shear fatigue tests showed that fatigue resistance of Sn-3.5Ag-XCu(X=0.75, 1.5), Sn-3.5Ag and Sn-0.7Cu alloys were more or less the same each other but much better than that of Bi containing alloys regardless of the surface finish layer used. In general, solder joints on the Au/Ni finish showed better fatigue resistance than those on the Cu finish. Cross-sectional fractography revealed microcracks nucleation inside of the interfacial IMC near the solder mask edge, more frequently on the Cu than the Au/Ni surface finish. Macro cracks followed the solder/IMC interface in the Bi containing alloys, while they propagated in the solder matrix in other alloys. It was ascribed to the Bi segregation at the solder/IMC interface and the solid solution hardening effect of Bi in the $\beta-Sn$ matrix.

• Journal of Conservation Science
• /
• v.25 no.2
• /
• pp.207-217
• /
• 2009

After replicating 10 bronze types such as Gyemija, Gyeongjaja, Eulhaeja, etc. before the Imjin war, we studied the change of microstructure from each casting process, method, and alloy ratio by Gyechukja replicated from "Donggukyeojiseungnam". We selected the average of compositions of Eulhaeja in the National Museum of Korea as the standard(Cu 86.7%, Sn: 9.7%, Pb: 2.3%) of bronze types, so we decided on the alloy's composition of Cu 87%, Sn 15%, Pb 8% added to 5% Sn and Pb contents because of evaporating the Sn and the Pb. Before replicating major metal types, we made master-alloy first, melting it again, and then replicated metal types. The composition of the 1'st replicated Gyechukja showed the range of Cu 85.81~87.63%, Sn 9.27~10.51%, Pb 3.05~3.19%. The 2'nd replicated Gyechukja made using the branch metal left after casting the 1st replica. The 2nd replicated Gyechukja showed the composition range of Cu 87.21~88.09%, Sn 9.06~9.36%, Pb 2.80~3.05%. This result decreases a little contents of Sn and Pb as compared with metal types of the 1st replica. However, it's almost the same as the Eulhaeja's average composition ratio in the National Museum of Korea. As a result of observing the microstructure of restored Gyechukja, it showed the dendrite structure of the typical casting structure and the segregation of Pb. There is no big difference of microstructure between the 1st and the 2nd restored metal types, even though the 2nd restored types partially decreases the eutectoid region in comparison with the 1st types. The systematic and scientific restoration experiment of metal types using Joseon period will be showed the casting method and alloy ratio, and this will be of great help to the study of restoration metal types in the future.

• Korean journal of communication and information
• /
• v.73
• /
• pp.56-91
• /
• 2015

This study examined changes in the independent production environment in the perspectives from flexible specialization of labor and media routines since January 2011, when comprehensive programming television channels (JTBC, MBN, Channel A, TV Chosun) emerged in Korea. In-depth interviews were conducted with thirteen individuals, including producers from independent production companies, broadcast writers, and individuals involved with these TV channels. The interview results indicated that a flexible specialization production system had been established by the comprehensive programming channels. This means that they were heavily dependent on independent producers, except in relations to their own news programs. Moreover, it was identified that the production of diverse programs could be difficult due to absurd contract practices such as those related to TV ratings and performance systems. Second, these channels have implemented some positive changes such as the payment of higher production costs and an incentive system, compared to terrestrial TV stations. However, the incentive system also helps to aggravate internal competition in the channel and also instigate contract competitions among independent companies, which can eventually result in the channels for holding exclusive rights to certain content and, hence, unfair business practices. Third, as a result of the newspaper and broadcast cross-owenership system of the comprehensive programming channels, hierarchical independent production practices can be established under the influence of newspaper proprietors and executives or managers who have previously worked for newspapers. Lastly, as a result of interviews with independent producers and individuals involved with the TV channels concerning the awareness of comprehensive programming channels, it could not be ascertained whether it is difficult to produce programs dealing with diverse items and genres, because programming autonomy has been distorted by capital or the advertisement market. In this circumstance, it is not surprising that some comprehensive programming channels mentioned that they prioritize profit and performance in programming. In conclusion, it is absolutely imperative that complementary and legal measures be implemented institutionally in order to redress the existing systematic dysfunctional routines in the independent productions of the comprehensive programming TV channels in Korea.

• Korean Journal of Heritage: History & Science
• /
• v.47 no.4
• /
• pp.224-243
• /
• 2014

This study examined the materials and casting technology(cast, alloy, etc.) used in the manufacturing of bronze artifacts based on old literature such as Yongjae Chonghwa, Cheongong Geamul, and The Korea Review. In the casting experiment for restoration of Sangpyong Tongbo, a bronze and brass mother coin mold was made using the sand mold casting method described in The Korea Review. The cast was comprised of the original mold plate frame, wooden frame, and molding sand. Depending on the material of the outer frame, which contains the molding sand, the original mold plate frame can be either a wooden frame or steel frame. For the molding sand, light yellow-colored sand of the Jeonbuk Iri region was used. Next, the composition of the mother alloy used in the restoration of Sangpyong Tongbo was studied. In consideration of the evaporation of tin and lead during actual restoration, the composition of Cu 60%, Zn 30%, and Pb 10% for brass as stated in The Korea Review was modified to Cu 60%, Zn 35%, and Pb 15%. For bronze, based on the composition of Cu 80%, Sn 6%, and Pb 14% used for Haedong Tongbo, the composition was set as Cu 80%, Sn 11%, and Pb 19%. The mother coin mold was restored by first creating a wooden father coin, making a cast from the wooden frame and basic steel frame, alloying, casting, and making a mother coin. Component analysis was conducted on the mother alloy of the restored Sangpyong Tongbo, and its primary and secondary casts. The bronze mother alloy saw a 5% increase in copper and 4% reduction in lead. The brass parent alloy had a 5% increase in copper, but a 4% and 12% decrease in lead and tin respectively. Analysis of the primary and secondary mother coin molds using an energy dispersive spectrometer showed that the bronze mother coin mold had a reduced amount of lead, while the brass mother coin mold had less tin. This can be explained by the evaporation of lead and tin in the melting of the primary mother coin mold. In addition, the ${\alpha}$-phase and lead particles were found in the mother alloy of bronze and brass, as well as the microstructure of the primary and secondary coin molds. Impurities such as Al and Si were observed only in the brass mother coin mold.