• Journal of Ocean Engineering and Technology
• /
• v.34 no.2
• /
• pp.147-154
• /
• 2020

Underwater acoustics that is the study of the phenomenon of underwater wave propagation and its interaction with boundaries, has mainly been applied to the fields of underwater communication, target detection, marine resources, marine environment, and underwater sound sources. Based on the scientific and engineering understanding of acoustic signals/data, recent studies combining traditional and data-driven machine learning methods have shown continuous progress. Machine learning, represented by deep learning, has shown unprecedented success in a variety of fields, owing to big data, graphical processor unit computing, and advances in algorithms. Although machine learning has not yet been implemented in every single field of underwater acoustics, it will be used more actively in the future in line with the ongoing development and overwhelming achievements of this method. To understand the research trends of machine learning applications in underwater acoustics, the general theoretical background of several related machine learning techniques is introduced in this paper.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.1 no.2
• /
• pp.169-181
• /
• 1997

Some association rules can have very high confidence in a sub-interval or a subrange of the domain, though not quite high confidence in the whole domain. In this paper, we define a ranged association rule, an association with high confidence worthy of special attention in a sub-domain, and further propose an efficient algorithm which finds out ranged association rules. The proposed algorithm is data-driven method in a sense that hypothetical subranges are built based on data distribution itself. In addition, to avoid redundant database scanning, we devise an effective in-memory data structure, that is collected through single database scanning. The simulation shows that the suggested algorithm has reliable performance at the acceptable time cost in actual application areas.

• Journal of The Korean Society of Agricultural Engineers
• /
• v.61 no.1
• /
• pp.31-44
• /
• 2019

The main objective of this research is to develop a hydroclimate drought index (HCDI) using the gridded climate data inputs in a Variable Infiltration Capacity (VIC) modeling platform. Typical drought indices, including, Standardized Precipitation Index (SPI), Standardized Precipitation Evapotranspiration Index (SPEI), and Self-calibrated Palmer Drought Severity Index (SC-PDSI) in South Korea are also used and compared. Inverse Distance Weighting (IDW) method is applied to create the gridded climate data from 56 ground weather stations using topographic information between weather stations and the respective grid cell ($12km{\times}12km$). R statistical software packages are used to visualize HCDI in Google Earth. Skill score (SS) are computed to evaluate the drought predictability based on water information derived from the observed reservoir storage and the ground weather stations. The study indicates that the proposed HCDI with the gridded climate data input is promising in the sense that it can help us to predict potential drought extents and to mitigate its impacts in a changing climate. The longer term drought prediction (e.g., 9 and 12 month) capability, in particular, shows higher SS so that it can be used for climate-driven future droughts.

• The Journal of Engineering Geology
• /
• v.19 no.4
• /
• pp.441-457
• /
• 2009

This study estimated ultimate load by the determination methods based on ultimate load, yield load and settlement using experimental data from static load tests that applied load to driven piles used in sandy grounds at home and overseas until failure appeared markedly. Estimated ultimate load was normalized with actually measured failure load, and was compared among the determination methods according to the characteristics of pile. In addition, I have identified to the determination methods suitable for estimating ultimate load, and reevaluated the safety factor when determining allowable load. From the results of this study were drawn conclusions as follows. Among ultimate loads estimated by the ultimate-load-based determination methods, the value interpreted by Chin's method tended to overestimate actual measurements, and B. Hansen 80% standard and the stability plot method were considered most reliable as their results were closest to actual measurements. According to the results of this study, in calculating the allowable load, if the safety factor to be applied to failing load obtained by the method of determining extreme load is converted to the safety factor applied to the Standards for Structure Foundation Design, a value larger than 3.0 should be applied except the B. Hansen 90% method, and a value larger than 2.0 should be applied in the methods of determining yield load. In addition, if the safety factor to be applied to load obtained by the settlement standard is converted based on safety factor 3.0 for extreme load, a value smaller than 3.0 should be applied to the total settlement standard and the net settlement standard.

• Journal of the Korea Society of Computer and Information
• /
• v.17 no.8
• /
• pp.81-90
• /
• 2012

For digital TV, the recommendation of u-health personalized service of semantic environment should be done after evaluating individual physical condition, illness and health condition. The existing recommendation method of u-health personalized service of semantic environment had low user satisfaction because its recommendation was dependent on ontology for analyzing significance. We propose the personalized service recommendation method based on Naive Bayesian Classifier for u-health service of semantic environment in digital TV. In accordance with the proposed method, the condition data is inferred by using ontology, and the transaction is saved. By applying naive bayesian classifier that uses preference information, the service is provided after inferring based on user preference information and transaction formed from ontology. The service inferred based on naive bayesian classifier shows higher precision and recall ratio of the contents recommendation rather than the existing method.

• Journal of KIISE:Information Networking
• /
• v.36 no.5
• /
• pp.413-424
• /
• 2009

Wireless sensor network is based on an event driven system. Sensor nodes collect the events in surrounding environment and the sensing data are relayed into a sink node. In particular, when events are detected, the data sensing periods are likely to be shorter to get the more correct information. However, this operation causes the traffic congestion on the sensor nodes located in a routing path. Since the traffic congestion generates the data queue overflows in sensor nodes, the important information about events could be missed. In addition, since the battery energy of sensor nodes exhausts quickly for treating the traffic congestion, the entire lifetime of wireless sensor networks would be abbreviated. In this paper, a new congestion control method is proposed on the basis of genetic algorithm. To apply genetic algorithm, the data traffic rate of each sensor node is utilized as a chromosome structure. The fitness function of genetic algorithm is designed from both the average and the standard deviation of the traffic rates of sensor nodes. Based on dominant gene sets, the proposed method selects the optimal data forwarding sensor nodes for relieving the traffic congestion. In experiments, when compared with other methods to handle the traffic congestion, the proposed method shows the efficient data transmissions due to much less queue overflows and supports the fair data transmission between all sensor nodes as possible. This result not only enhances the reliability of data transmission but also distributes the energy consumptions across the network. It contributes directly to the extension of total lifetime of wireless sensor networks.

• Journal of the Korean Geotechnical Society
• /
• v.25 no.12
• /
• pp.107-118
• /
• 2009

As CPT penetration tends to show a similar behavior to that of pile driving, a number of methods for estimating the toe bearing capacity of piles based on CPT data have been proposed. To evaluate the applicability of the methods in this country, a total of 172 dynamic load tests data on PHC piles and 82 CPT data at a site in the Nakdong River estuary were collected. A specific four-step procedure was adopted for the selection of the reliable data, and statistical techniques were then applied to the analysis of the applicability. The results indicated that among a total of 10 CPT-based methods applied, the best one is the Aoki method (1975), followed by the LCPC (1982), ICP (2005) methods and others.

• Journal of Korean Society for Quality Management
• /
• v.49 no.3
• /
• pp.421-431
• /
• 2021

Purpose: For improving outgoing quality, this study presents a novel sampling framework based on predictive analytics. Methods: The proposed framework is composed of three steps. The first step is the variable selection. The knowledge-based and data-driven approaches are employed to select important variables. The second step is the model learning. In this step, we consider the supervised classification methods, the anomaly detection methods, and the rule-based methods. The applying model is the third step. This step includes the all processes to be enabled on real-time prediction. Each prediction model classifies a product as a target sample or random sample. Thereafter intensive quality inspections are executed on the specified target samples. Results: The inspection data of three Samsung products (mobile, TV, refrigerator) are used to check functional defects in the product by utilizing the proposed method. The results demonstrate that using target sampling is more effective and efficient than random sampling. Conclusion: The results of this paper show that the proposed method can efficiently detect products that have the possibilities of user's defect in the lot. Additionally our study can guide practitioners on how to easily detect defective products using stratified sampling

• Nuclear Engineering and Technology
• /
• v.54 no.1
• /
• pp.36-48
• /
• 2022

As a valid numerical method to obtain a high-resolution result of a flow field, computational fluid dynamics (CFD) have been widely used to study coolant flow and heat transfer characteristics in fuel rod bundles. However, the time-consuming, iterative calculation of Navier-Stokes equations makes CFD unsuitable for the scenarios that require efficient simulation such as sensitivity analysis and uncertainty quantification. To solve this problem, a reduced-order model (ROM) based on proper orthogonal decomposition (POD) and machine learning (ML) is proposed to simulate the flow field efficiently. Firstly, a validated CFD model to output the flow field data set of the rod bundle is established. Secondly, based on the POD method, the modes and corresponding coefficients of the flow field were extracted. Then, an deep feed-forward neural network, due to its efficiency in approximating arbitrary functions and its ability to handle high-dimensional and strong nonlinear problems, is selected to build a model that maps the non-linear relationship between the mode coefficients and the boundary conditions. A trained surrogate model for modes coefficients prediction is obtained after a certain number of training iterations. Finally, the flow field is reconstructed by combining the product of the POD basis and coefficients. Based on the test dataset, an evaluation of the ROM is carried out. The evaluation results show that the proposed POD-ROM accurately describe the flow status of the fluid field in rod bundles with high resolution in only a few milliseconds.

• Journal of Wind Energy
• /
• v.14 no.3
• /
• pp.34-42
• /
• 2023

The reduction of operation and maintenance (O&M) costs is a critical factor in determining the competitiveness of wind energy. Predictive maintenance based on the estimation of remaining useful life (RUL) is a key technology to reduce logistic costs and increase the availability of wind turbines. Although a mechanical component usually has sudden changes during operation, most RUL estimation methods use the trend of a state index over the whole operation period. Therefore, overestimation of RUL causes confusion in O&M plans and reduces the effect of predictive maintenance. In this paper, two RUL estimation methods (load based and data driven) are proposed for the bearings of a wind turbine with the results of trend classification, which differentiates constant and increasing states of the state index. The proposed estimation method is applied to a bearing degradation test, which shows a conservative estimation of RUL.