• Journal of the Korea Society of Computer and Information
• /
• v.25 no.4
• /
• pp.19-27
• /
• 2020

In this paper, we propose a machine learning method for diagnosing the failure of a gas pressure regulator. Originally, when implementing a machine learning model for detecting abnormal operation of a facility, it is common to install sensors to collect data. However, failure of a gas pressure regulator can lead to fatal safety problems, so that installing an additional sensor on a gas pressure regulator is not simple. In this paper, we propose various machine learning approach for diagnosing the abnormal operation of a gas pressure regulator with only the flow rate and gas pressure data collected from a gas pressure regulator itself. Since the fault data of a gas pressure regulator is not enough, the model is trained in all classes by applying the over-sampling method. The classification model was implemented using Gradient boosting, 1D Convolutional Neural Networks, and LSTM algorithm, and gradient boosting model showed the best performance among classification models with 99.975% accuracy.

• Annual Conference of KIPS
• /
• 2022.11a
• /
• pp.1047-1049
• /
• 2022

스마트 컨테이너 제어를 위해서는 컨테이너 내부에 센서가 필요하나, 센서의 개수가 증가하면 비용 및 시스템 부하가 증가한다. 본 연구에서는 CFD(Computational Fluid Dynamics)를 이용하여 얻은 컨테이너 내부 온도 데이터와 센서 위치 최적화 알고리즘을 이용하여 컨테이너 내부 모니터링을 위한 최적의 센서 위치 결정 방법론을 제시한다. CFD 상용 SW로 컨테이너 내·외부 상황을 가정하여 내부 온도 데이터를 추출하고, 이를 바탕으로 내부 상태를 대표하는 공간들을 구분한다. 컨테이너 내벽에 부착된 센서가 탐지할 수 있는 능력을 탐지 거리 및 각도의 수식들로 나타내어 각 수식을 조합하여 센서의 탐지 능력을 수치화하고, 이 수치에 따라 균등하게 분포된 센서 위치 후보군 중, 선별된 공간을 탐지하는 센서 위치를 최적화하여 효율적인 컨테이너 제어를 위한 여건을 마련한다.

• Journal of Internet Computing and Services
• /
• v.20 no.1
• /
• pp.77-86
• /
• 2019

As the number of single-person households increases, it is not easy to ask for help alone if a single-person household is severely injured in the home. This paper detects abnormal event when members of a single household in the home are seriously injured. It proposes an vision detection algorithm that analyzes and recognizes patterns through videos that are collected based on home CCTV. And proposes audio detection algorithms that analyze and recognize patterns of sound that occur in households based on Smartphones. If only each algorithm is used, shortcomings exist and it is difficult to detect situations such as serious injuries in a wide area. So I propose a fusion method that effectively combines the two algorithms. The performance of the detection algorithm and the precise detection performance of the proposed fusion method were evaluated, respectively.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.26 no.3
• /
• pp.246-252
• /
• 2016

Since thermal power plant (TPP) equipment is operated under very high pressure and temperature, failures of the equipment give rise to severe losses of life and property. To prevent the losses, fault detection method is, therefore, absolutely necessary to identify abnormal operating conditions of the equipment in advance. In this paper, we present Mahalanobis distance (MD) based fault detection method for steam boiler tube in TPP. In the MD-based method, it is supposed that abnormal data samples are far away from normal samples. Using multivariate samples collected from normal target system, mean vector and covariance matrix are calculated and threshold value of MD is decided. In a test phase, after calculating the MDs between the mean vector and test samples, alarm signals occur if the MDs exceed the predefined threshold. To demonstrate the performance, a failure case due to boiler tube leakage in 200MW TPP is employed. The experimental results show that the presented method can perform early detection of boiler tube leakage successfully.

• Journal of the Korean Society for Marine Environment & Energy
• /
• v.16 no.1
• /
• pp.1-8
• /
• 2013

This study focuses on a ship detection and tracking method using Frequency Modulated Continuous Wave (FMCW) radar used for horizontal surveillance. In general, FMCW radar can play an important role in maritime surveillance, because it has many advantages such as low warm-up time, low power consumption, and its all weather performance. In this paper, we introduce an effective method for data and signal processing of ship's detecting and tracking using the X-band radar. Ships information was extracted using an image-based processing method such as the land masking and morphological filtering with a threshold for a cycle data merged from raw data (spoke data). After that, ships was tracked using search-window that is ship's expected rectangle area in the next frame considering expected maximum speed (19 kts) and interval time (5 sec). By using this method, the tracking results for most of the moving object tracking was successful and those results were compared with AIS (Automatic Identification System) for ships position. Therefore, it can be said that the practical application of this detection and tracking method using FMCW radar improve the maritime safety as well as expand the surveillance coverage cost-effectively. Algorithm improvements are required for an enhancement of small ship detection and tracking technique in the future.

• Journal of the Korean Data and Information Science Society
• /
• v.26 no.3
• /
• pp.639-652
• /
• 2015

In today's manufacturing environment, the process data can be easily measured and transferred to a computer for analysis in a real-time mode. As a result, it is possible to monitor several correlated quality variables simultaneously. Various multivariate statistical process control (MSPC) procedures have been presented to detect an out-of-control event. Although the classical MSPC procedures give the out-of-control signal, it is difficult to determine which variable has caused the signal. In order to solve this problem, data mining and machine learning techniques can be considered. In this paper, we applied the technique of decision tree learning to the MSPC, and we did simulation for MSPC procedures to monitor the bivariate normal process means. The results of simulation show that the overall performance of the MSPC procedure using decision tree learning technique is similar for several values of correlation coefficient, and the accurate classification rates for out-of-control are different depending on the values of correlation coefficient and the shift magnitude. The introduced procedure has the advantage that it provides the information about assignable causes, which can be required by practitioners.

• Review of KIISC
• /
• v.15 no.3
• /
• pp.69-75
• /
• 2005

특정 또는 불특정 네트워크를 공격 대상으로 하는 인터넷 웜, 분산서비스거부 공격 등의 출현과 가공할 파괴력으로 인해 네트워크에 대한 보안 요구가 점차 증가하고 있다. 침입탐지시스템 등의 네트워크 보안 솔루션은 네트워크 공격에 대한 감시 및 차단 등의 기능을 제공하며 기술적인 진화를 거듭하고 있지만, 근본적인 한계로써 국부 감시로 인한 불명확성과 오탐에 의한 보안경보$^{1)}$ 플러딩 등이 지적되고 있다. 특히 보안경보의 플러딩 현상은 네트워크 보안 상태를 정확하게 판단하는 것을 방해함으로써 조기경보체제의 구축을 어렵게 하는 요인이 되고 있다. 최근 이러한 부분을 극복하기 위해 보안경보 간의 상호 연관성 분석에 대한 연구가 활발해 지고 있다. 본 논문에서는 보안경보에 대한 연관성 분석 동향에 대해서 논의한다. 또한 보안경보의 집단화(aggregation)를 이용한 네트워크 공격 상황 분석방안에 대해서도 논의한다. 보안경보의 집단화를 이용한 공격 상황 분석은 엄청나게 발생하는 보안경보로부터 조기경보를 위한 공격 정보의 판별과 광역 네트워크상에서 이상 현상의 탐지를 가능하게 한다. 이와 더불어 현재 ETRI에서 개발 중에 있는 네트워크 공격 상황 분석기인 NASA(Network Attack Situation Analyzer)에 대해서도 간략히 소개한다.

• Annual Conference of KIPS
• /
• 2012.11a
• /
• pp.523-526
• /
• 2012

영상의 위변조를 탐지하거나 무결성을 인증하기 위해서는 가역 워터마킹 기법은 유용하다. 기존 워터 마킹 연구들은 원본 복원이 불가능하였으나, 가역 워터마킹은 워터마크를 검출한 후, 아무런 손상없이 영상을 원본 상태로 복원할 수 있는 방법이다. 본 논문에서는 차이값 히스토그램에 기반한 가역 워터 마킹을 통해 위변조된 영역을 탐지하는 블록단위 인증 알고리즘을 제안한다. 먼저, 영상 각 블록에 대하여 영상의 특징값을 추출하고, 사용자의 정보와 결합하여 인증 코드를 생성한다. 생성된 인증코드는 가역 워터마킹을 통하여 콘텐츠 자체에 직접 삽입한다. 영상의 인증을 위해서는 추출된 인증코드와 새로 생성된 인증코드의 비교를 수행한다. 다양한 영상들에 대하여 비교 분석하였고, 그 결과 제안한 알고리즘은 완전한 가역성과 함께 낮은 왜곡을 유지하면서도 97% 이상 인증률을 얻을 수 있었다.

• The KIPS Transactions:PartC
• /
• v.14C no.3 s.113
• /
• pp.199-208
• /
• 2007

In a ubiquitous network environment, detecting the leakage of personal information is very important. The leakage of personal information might cause severe problem such as impersonation, cyber criminal and personal privacy violation. In this paper, we have proposed a detection method of personal information leakage based on network traffic characteristics. The experimental results indicate that the traffic character of a real campus network shows the self-similarity and Proposed method can detect the anomaly of leakage of personal information by malicious code.

• Journal of IKEEE
• /
• v.22 no.3
• /
• pp.532-538
• /
• 2018

This paper proposes an algorithm for signal processing which is used in pulse induction metal mine detectors. The detection power can be obtained from magnetic variation on the search coil. The calibration data should be made when there is no target because the detection power is difference between with and without a target. And it is also updated periodically because of surrounding various noises. Lastly, we keep a watch on the signal slope to identify exact position and signal power of mine detection.