• Journal of Industrial Technology
• /
• v.28 no.B
• /
• pp.39-45
• /
• 2008

Temperature variation according to space and time on the inner parts of engineering constructions(e.g.: dam, slope) can be a basic information for diagnosing their safety problem. In general, as constructions become superannuated, structural deformation(e.g.: cracks, defects) could be occurred by various factors. Seepage or leakage of water through these cracks or defects in old dams will directly cause temperature anomaly. Groundwater level also can be easily observed by abrupt change of temperature on the level. This study shows that the position of seepage or leakage in dam body can be detected by multi-channel temperature monitoring using thermal line sensor. For this, diverse temperature monitoring experiments for a leakage physical model were performed in the laboratory. In field application of an old earth fill dam, temperature variations for water depth and for inner parts of boreholes located at downstream slope were measured. Temperature monitoring results for a long time at the bottom of downstream slope of the dam showed the possibility that temperature monitoring can provide the synthetic information about flowing path and quantity of seepage of leakage in dam body.

• Geomechanics and Engineering
• /
• v.25 no.1
• /
• pp.59-73
• /
• 2021

There frequently exists inadequacy regarding the number of boreholes installed along tunnel alignment. While geophysical imaging techniques are available for pre-tunnelling geological characterization, they aim to detect specific object (e.g., water body and karst cave). There remains great motivation for the industry to develop a real-time identification technology relating complex geological conditions with the existing tunnelling parameters. This study explores the potential for the use of machine learning-based data driven approaches to identify the change in geology during tunnel excavation. Further, the feasibility for machine learning-based anomaly detection approaches to detect the development of clayey clogging is also assessed. The results of an application of the machine learning-based approaches to Xi'an Metro line 4 are presented in this paper where two tunnels buried in the water-rich sandy soils at depths of 12-14 m are excavated using a 6.288 m diameter EPB shield machine. A reasonable agreement with the measurements verifies their applicability towards widening the application horizon of machine learning-based approaches.

• Journal of Korea Society of Industrial Information Systems
• /
• v.25 no.6
• /
• pp.1-13
• /
• 2020

In this paper, the author proposed a nuclear power plant (NPP) instrumentation signal faults identification algorithm. A variational autoencoder (VAE)-based model is trained by using only normal dataset as same as existing anomaly detection method, and trained model predicts which signal within the entire signal set is anomalous. Classification of anomalous signals is performed based on the reconstruction error for each kind of signal and partial derivatives of reconstruction error with respect to the specific part of an input. Simulation was conducted to acquire the data for the experiments. Through the experiments, it was identified that the proposed signal fault identification method can specify the anomalous signals within acceptable range of error.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• v.26 no.4
• /
• pp.280-295
• /
• 2022

This paper contains an introduction to industrial problems, solutions, and results conducted with the Korea Association of Machinery Industry. The client company commissioned the problem of upgrading the method of identifying global supply risky items. Accordingly, the factors affecting the supply and demand of imported items in the global supply chain were identified and the method of selecting risky items was studied and delivered. Through research and discussions with the client companies, it is confirmed that the most suitable factors for identifying global supply risky items are 'import size', 'import dependence', and 'trend abnormality'. The meaning of each indicator is introduced, and risky items are selected using export/import data until October 2022. Through this paper, it is expected that countries and companies will be able to identify global supply risky items in advance and prepare for risks in the new normal situation: the economic situation caused by infectious diseases such as the COVID-19 pandemic; and the export/import regulation due to geopolitical problems. The client company will include in his report, the method presented in this paper and the risky items selected by the method.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2024.01a
• /
• pp.299-300
• /
• 2024

첨단 기술의 발전과 함께 지능형 운전자 보조 시스템의 성능 및 교통 시스템 체계가 고도화됨에 따라 전반적인 교통사고 발생 건수는 줄어드는 추세지만 대한민국의 교통사고 발생 빈도는 아직 OECD 평균 대비 높은 실정이다. 특히, 2020년 경제 협력 개발 기구(OECD) 통계에 따르면 대한민국의 인구 10만 명당 교통사고 사망자 수는 회원국 36개 중 29위로 매우 높은 축에 속한다. 따라서, 본 논문에서는 교통사고 발생률을 낮추는 데 도움을 줄 수 있는 주행 이상 패턴 감지 시스템을 제안한다. 제안한 방법에서는 실시간 영상 분석을 통해 신호등 및 차선을 인식함과 동시 차량 내부 진단 데이터에 대한 시계열 분석을 기반으로 운전자의 운전 패턴을 분석한 후 평소와 다른 이상 징후를 발견하면 운전자에게 경고 알림을 제공하여 위험한 상황을 회피할 수 있도록 지원한다.

• The Sea:JOURNAL OF THE KOREAN SOCIETY OF OCEANOGRAPHY
• /
• v.26 no.4
• /
• pp.307-326
• /
• 2021

Real-time sea level observations from tide gauges include missing and erroneous values. Classification as abnormal values can be done for the latter by the quality control procedure. Although the 3𝜎 (three standard deviations) rule has been applied in general to eliminate them, it is difficult to apply it to the sea-level data where extreme values can exist due to weather events, etc., or where erroneous values can exist even within the 3𝜎 range. An artificial intelligence model set designed in this study consists of non-annotated recurrent neural networks and ensemble techniques that do not require pre-labeling of the abnormal values. The developed model can identify an erroneous value less than 20 minutes of tide gauge recording an abnormal sea level. The validated model well separates normal and abnormal values during normal times and weather events. It was also confirmed that abnormal values can be detected even in the period of years when the sea level data have not been used for training. The artificial neural network algorithm utilized in this study is not limited to the coastal sea level, and hence it can be extended to the detection model of erroneous values in various oceanic and atmospheric data.

• Journal of the Korean Society of Industry Convergence
• /
• v.26 no.1
• /
• pp.113-119
• /
• 2023

In this paper, we propose an optimization of a pose estimation deep learning model for recognition of abnormal behavior in unmanned stores using radio frequencies. The radio frequency use millimeter wave in the 30 GHz to 300 GHz band. Due to the short wavelength and strong straightness, it is a frequency with less grayness and less interference due to radio absorption on the object. A millimeter wave radar is used to solve the problem of personal information infringement that may occur in conventional CCTV image-based pose estimation. Deep learning-based pose estimation models generally use convolution neural networks. The convolution neural network is a combination of convolution layers and pooling layers of different types, and there are many cases of convolution filter size, number, and convolution operations, and more cases of combining components. Therefore, it is difficult to find the structure and components of the optimal posture estimation model for input data. Compared with conventional millimeter wave-based posture estimation studies, it is possible to explore the structure and components of the optimal posture estimation model for input data using genetic algorithms, and the performance of optimizing the proposed posture estimation model is excellent. Data are collected for actual unmanned stores, and point cloud data and three-dimensional keypoint information of Kinect Azure are collected using millimeter wave radar for collapse and property damage occurring in unmanned stores. As a result of the experiment, it was confirmed that the error was moored compared to the conventional posture estimation model.

• Journal of KIISE:Information Networking
• /
• v.33 no.3
• /
• pp.201-217
• /
• 2006

In this paper, we collected the physical traces from high speed Internet backbone traffic and analyze the various characteristics of the underlying packet traces. Particularly, our work is focused on analyzing the characteristics of an anomalous traffic. It is found that in our data, the anomalous traffic is caused by UDP session traffic and we determined that it was one of the Denial of Service attacks. In this work, we adopted the unsupervised machine learning algorithm to classify the network flows. We apply the k-means clustering algorithm to train the learner. Via the Cramer-Yon-Misses test, we confirmed that the proposed classification method which is able to detect anomalous traffic within 1 second can accurately predict the class of a flow and can be effectively used in determining the anomalous flows.

• Journal of Satellite, Information and Communications
• /
• v.12 no.3
• /
• pp.35-40
• /
• 2017

Out-of-limit (OOL) alarm method that is threshold checking of telemetry value is widely used for the satellites fault diagnosis and health monitoring. However, it requires engineering knowledge and effort to define delicate threshold value and has limitations that anomalous behaviors within the defined limits can't be detected. In this paper, we propose a satellite anomalous behavior detection system through fuzzy model that is composed by important statistical feature selected by rough-set theory. Not pre-defined anomaly is detected because only normal state data is used for fuzzy model. Also, anomalous behavior within the threshold limit is detected by using statistic feature that can be collected without engineering knowledge. The proposed system successfully detected non-ordinary state for battery temperature telemetry.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.8 no.4
• /
• pp.1307-1323
• /
• 2014

With the exhaustion of global IPv4 addresses, IPv6 technologies have attracted increasing attentions, and have been deployed widely. Meanwhile, new applications running over IPv6 networks will change the traditional traffic characteristics obtained from IPv4 networks. Traditional models obtained from IPv4 cannot be used for IPv6 network monitoring directly and there is a need to investigate those changes. In this paper, we explore the flow features of IPv6 traffic and compare its difference with that of IPv4 traffic from flow level. Firstly, we analyze the differences of the general flow statistical characteristics and users' behavior between IPv4 and IPv6 networks. We find that there are more elephant flows in IPv6, which is critical for traffic engineering. Secondly, we find that there exist many one-way flows both in the IPv4 and IPv6 traffic, which are important information sources for abnormal behavior detection. Finally, in light of the challenges of analyzing massive data of large-scale network monitoring, we propose a group flow model which can greatly reduce the number of flows while capturing the primary traffic features, and perform a comparative measurement analysis of group users' behavior dynamic characteristics. We find there are less sharp changes caused by abnormity compared with IPv4, which shows there are less large-scale malicious activities in IPv6 currently. All the evaluation experiments are carried out based on the traffic traces collected from the Northwest Regional Center of CERNET (China Education and Research Network), and the results reveal the detailed flow characteristics of IPv6, which are useful for traffic management and anomaly detection in IPv6.