• Korean Journal of Metals and Materials
• /
• v.46 no.5
• /
• pp.304-309
• /
• 2008

Magnesium alloys containing $Mg_2Si$ particles, as a promising cheap heat-resistant magnesium alloy for automobile power train parts applications, are attracting more attention of both material scientists and design engineers. Modification of the Chinese script shape $Mg_2Si$ particle is a key for using this alloy in sand or permanent mould casting. In the present work, the modification effect of Sr and Sb on the corrosion properties of the Mg-5Al-2Si alloy was investigated. Sr or Sb addition promoted the formation of fine polygonal shape $Mg_2Si$ particles by providing the nucleation sites. Sr was more effective element than Sb for shape modification of Chinese script shape $Mg_2Si$. Such improved microstructure of the modified alloy resulted in large improvement in corrosion resistance as compared to unmodified Mg-5Al-2Si alloy.

• Nuclear Engineering and Technology
• /
• v.52 no.12
• /
• pp.2709-2716
• /
• 2020

Compositions of large nuclear cores (e.g. boiling water reactors) are highly heterogeneous in terms of fuel composition, control rod insertions and flow regimes. For this reason, they usually lack high order of symmetry (e.g. 1/4, 1/8) making it difficult to estimate their neutronic parameters for large spaces of possible loading patterns. A detailed hyperparameter optimization technique (a combination of manual and Gaussian process search) is used to train and optimize deep neural networks for the prediction of three neutronic parameters for the Ringhals-1 BWR unit: power peaking factors (PPF), control rod bank level, and cycle length. Simulation data is generated based on half-symmetry using PARCS core simulator by shuffling a total of 196 assemblies. The results demonstrate a promising performance by the deep networks as acceptable mean absolute error values are found for the global maximum PPF (~0.2) and for the radially and axially averaged PPF (~0.05). The mean difference between targets and predictions for the control rod level is about 5% insertion depth. Lastly, cycle length labels are predicted with 82% accuracy. The results also demonstrate that 10,000 samples are adequate to capture about 80% of the high-dimensional space, with minor improvements found for larger number of samples. The promising findings of this work prove the ability of deep neural networks to resolve high dimensionality issues of large cores in the nuclear area.

• Journal of the Korea Safety Management & Science
• /
• v.20 no.4
• /
• pp.1-6
• /
• 2018

Recently, bicycle has means of effective healthy transportation, and riding the bicycles is considered as popular recreational and sporting activities. Also, the saddle, steering system, driving device and braking device are researched briskly because of consumer's need for driving performance and comfort. Especially, the importance of a cassette responsible for transmission function by transmitting power to the drive shaft through the chain is very focused. The writer conducted structural analysis for the sprocket of each level using the ANSYS widely used for the analysis. Speed shifting performance was enhanced by minimization / simplification of shifting point through a sort of tooth profile of the cassette. By partitioning a clear value type and other shifting point, it has been modified to enable smooth speed-shifting. In addition, as titanium precision forming process, this study studied the molding technique by blanking and dies forging for mass production of the cassette. so it could be expected that the entire drive train would utilize that in the future. The stamping process capability for thin materials for the mass production of the sprockets is applicable to producing automobile parts, so lightweight component production is likely to be possible through that, for the safety of driving.

• Computers and Concrete
• /
• v.31 no.4
• /
• pp.327-335
• /
• 2023

The most popular building material, concrete, is intrinsically linked to the advancement of humanity. Due to the ever-increasing complexity of cementitious systems, concrete formulation for desired qualities remains a difficult undertaking despite conceptual and methodological advancement in the field of concrete science. Recognising the significant pollution caused by the traditional cement industry, construction of civil engineering structures has been carried out successfully using Geopolymer Concrete (GPC), also known as High Performance Concrete (HPC). These are concretes formed by the reaction of inorganic materials with a high content of Silicon and Aluminium (Pozzolans) with alkalis to achieve cementitious properties. These supplementary cementitious materials include Ground Granulated Blast Furnace Slag (GGBFS), a waste material generated in the steel manufacturing industry; Fly Ash, which is a fine waste product produced by coal-fired power stations and Silica Fume, a by-product of producing silicon metal or ferrosilicon alloys. This result demonstrated that GPC/HPC can be utilised as a substitute for traditional Portland cement-based concrete, resulting in improvements in concrete properties in addition to environmental and economic benefits. This study explores utilising experimental data to train artificial neural networks, which are then used to determine the effect of supplementary cementitious material replacement, namely fly ash, Ground Granulated Blast Furnace Slag (GGBFS) and silica fume, on the compressive strength, tensile strength, and modulus of elasticity of concrete and to predict these values accordingly.

• Nuclear Engineering and Technology
• /
• v.55 no.8
• /
• pp.2747-2756
• /
• 2023

Reinforced Concrete (RC) shear walls are one of the civil structures in nuclear power plants to resist lateral loads such as earthquakes and wind loads effectively. Risk-informed and performance-based regulation in the nuclear industry requires considering possible accidents and determining desirable performance on structures. As a result, rather than predicting only the ultimate capacity of structures, the prediction of performances on structures depending on different damage states or various accident scenarios have increasingly needed. This study aims to develop machine-learning models predicting drifts of the RC shear walls according to the damage limit states. The damage limit states are divided into four categories: the onset of cracking, yielding of rebars, crushing of concrete, and structural failure. The data on the drift of shear walls at each damage state are collected from the existing studies, and four regression machine-learning models are used to train the datasets. In addition, the bagging ensemble method is applied to improve the accuracy of the individual machine-learning models. The developed models are to predict the drifts of shear walls consisting of various cross-sections based on designated damage limit states in advance and help to determine the repairing methods according to damage levels to shear walls.

• Journal of the Korea Society for Simulation
• /
• v.19 no.2
• /
• pp.107-118
• /
• 2010

Automotive manufacturing is a complex task that requires the production and assembly of thousands of different components or parts. The engine and the transmission are the major components that constitute a power train system. Although manufacturing processes of an engine are similar, the layouts of the manufacturing lines are different from factory to factory. It is due to the different design concept that how to combine the serial and parallel structures. In this paper, three engine lines of different factories are introduced, and the simulation technology is used to make the performance analysis for different design concepts.

• The Journal of the Acoustical Society of Korea
• /
• v.41 no.6
• /
• pp.629-636
• /
• 2022

We present a preprocessing method for an underwater target detection model based on a convolutional neural network. The acoustic characteristics of the ship show ambiguous expression due to the strong signal power of the low frequency. To solve this problem, we combine feature preprocessing methods with various feature scaling methods and spectrogram methods. Define a simple convolutional neural network model and train it to measure preprocessing performance. Through experiment, we found that the combination of log Mel-spectrogram and standardization and robust scaling methods gave the best classification performance.

• International conference on construction engineering and project management
• /
• 2020.12a
• /
• pp.387-396
• /
• 2020

Wearable biosensors have the potential to non-invasively and continuously monitor seniors' stress in their daily interaction with the urban environment, thereby enabling to address the stress and ultimately advance their outdoor mobility. However, current wearable biosensor-based stress detection methods have several drawbacks in field application due to their dependence on batch-learning algorithms. First, these methods train a single classifier, which might not account for multiple subjects' different physiological reactivity to stress. Second, they require a great deal of computational power to store and reuse all previous data for updating the signle classifier. To address this issue, we tested the feasibility of online multi-task learning (OMTL) algorithms to identify multiple seniors' stress from electrodermal activity (EDA) collected by a wristband-type biosensor in a daily trip setting. As a result, OMTL algorithms showed the higher test accuracy (75.7%, 76.2%, and 71.2%) than a batch-learning algorithm (64.8%). This finding demonstrates that the OMTL algorithms can strengthen the field applicability of the wearable biosensor-based stress detection, thereby contributing to better understanding the seniors' stress in the urban environment and ultimately advancing their mobility.

• Journal of the Semiconductor & Display Technology
• /
• v.22 no.4
• /
• pp.13-18
• /
• 2023

A subway station is a prominent multi-purpose facility where the quantitative management of fine dust, generated by various factors, is conducted. Recently, eco-friendly air purification methods using air-purifying plants are being discussed, with the focus on biofiltration through vegetation. Previous research in this field has confirmed the reduction effects of transition metals such as Fe, which have been identified as harmful to human health. This study aimed to identify the sources of fine dust dispersion within subway stations and derive an efficient operational strategy for air-purifying plants that takes into account the behavior characteristics of fine dust within multi-purpose facilities. The experiment monitored regional fine dust levels through IAQ stations established based on prior research. Also, the data was analyzed through time-series and correlation analyses by linking it with passenger counts at subway stations and the frequency of train stops. Furthermore, to consider energy efficiency, we conducted component-specific power consumption monitoring. Through this study, we were able to derive the optimal operational strategy for air-purifying plants based on time-series comprehensive analysis data and confirm significant energy efficiency.

• Nuclear Engineering and Technology
• /
• v.55 no.12
• /
• pp.4607-4616
• /
• 2023

We implement machine learning regression models to predict peak pressures of primary and secondary systems, a major safety concern in Loss Of Condenser Vacuum (LOCV) accident. We selected the Multi-dimensional Analysis of Reactor Safety-KINS standard (MARS-KS) code to analyze the LOCV accident, and the reference plant is the Korean Optimized Power Reactor 1000MWe (OPR1000). eXtreme Gradient Boosting (XGBoost) is selected as a machine learning tool. The MARS-KS code is used to generate LOCV accident data and the data is applied to train the machine learning model. Hyperparameter optimization is performed using a simulated annealing. The randomly generated combination of initial conditions within the operating range is put into the input of the XGBoost model to predict the peak pressure. These initial conditions that cause peak pressure with MARS-KS generate the results. After such a process, the error between the predicted value and the code output is calculated. Uncertainty about the machine learning model is also calculated to verify the model accuracy. The machine learning model presented in this paper successfully identifies a combination of initial conditions that produce a more conservative peak pressure than the values calculated with existing methodologies.