• Journal of Electrical Engineering and Technology
• /
• v.10 no.3
• /
• pp.719-728
• /
• 2015

This paper presents the results of field tests on Voltage Management System (VMS) using hybrid voltage control, which utilizes coordinated controls of various reactive power resources such as generators, FACTS and switched shunt devices to regulate the pilot bus voltage in a voltage control area. It also includes the results of performance test on RTDS-based test bed in order to validate the VMS before installing it in Jeju power system. The main purpose of the system is adequately to regulate the reactive power reserve of key generators in a normal condition with coordination of discrete shunt devices such as condensers and reactors so that the reserves can avoid voltage collapse in emergency state in Jeju system. Field tests in the automatic mode of VMS operation are included in steady-states and transient states. Finally, by the successful operation of VMS in Jeju power system, the VMS is proved to effectively control system voltage profiles in steady-state condition, increase system MVAR reserves and improve system reliability for pre- and post-contingency.

• Journal of Korea Technology Innovation Society
• /
• v.20 no.2
• /
• pp.271-291
• /
• 2017

This study measures and compares the performance of research and development (R&D) programs in government-funded research institutes (GRIs) in terms of qualitative and quantitative approaches to find out strategic insights for improving performance evaluation policy. In particular, we adopt the evaluation results from the real data of K institute in 2015 for a qualitative evaluation and the results of data envelopment analysis (DEA) for a quantitative evaluation. Comparative analysis of the R&D performance of 14 programs finds that the difference between the evaluation results of qualitative and quantitative approaches is significant. From this finding, we suggest several strategic directions to complement two approaches each other.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.40 no.3
• /
• pp.327-336
• /
• 2016

Manufacturers of aircraft engines have introduced computer-aided manufacturing (CAM) software to operate and control computerized numerical control (CNC) machine tools. However, the generation of a CAM model is a time consuming and error-prone task since machining procedure and operational details are manually defined. For the automatic generation of a CAM model, feature recognition techniques have been widely studied. However, their recognition coverage is limited so that complex shapes such as a jet engine cannot be fully developed. This study presents a novel approach to quickly generate a CAM model from a CAD model using shape search techniques. Once an operator sets a machining operation as a reference operation, the same shapes as the shapes referenced by the operation are searched. The reference operation is copied to the positions of the searched shapes. The proposed method was verified through experiments with a jet engine compressor case.

• International journal of advanced smart convergence
• /
• v.8 no.1
• /
• pp.24-34
• /
• 2019

In this research, a practical deep learning framework to differentiate the lesions and nodules in breast acquired with ultrasound imaging has been proposed. 7408 ultrasound breast images of 5151 patient cases were collected. All cases were biopsy proven and lesions were semi-automatically segmented. To compensate for the shift caused in the segmentation, the boundaries of each lesion were drawn using Fully Convolutional Networks(FCN) segmentation method based on the radiologist's specified point. The data set consists of 4254 benign and 3154 malignant lesions. In 7408 ultrasound breast images, the number of training images is 6579, and the number of test images is 829. The margin between the boundary of each lesion and the boundary of the image itself varied for training image augmentation. The training images were augmented by varying the margin between the boundary of each lesion and the boundary of the image itself. The images were processed through histogram equalization, image cropping, and margin augmentation. The networks trained on the data with augmentation and the data without augmentation all had AUC over 0.95. The network exhibited about 90% accuracy, 0.86 sensitivity and 0.95 specificity. Although the proposed framework still requires to point to the location of the target ROI with the help of radiologists, the result of the suggested framework showed promising results. It supports human radiologist to give successful performance and helps to create a fluent diagnostic workflow that meets the fundamental purpose of CADx.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.48 no.12
• /
• pp.953-960
• /
• 2020

This paper presents an optimization of gear web design used in a main gear train of an engine reduction gearbox for a rotorcraft. The optimization involves the minimization of a total weight, transmission error, misalignment, and face load distribution factor. In particular, three design variables such as a gear web thickness, location of rim-web connection, and location of shaft-web connection were set as design parameters. In the optimization process, web, rim and shaft of gears were converted from the 3D CAD geometry model to the finite element model, and then provided as input to the gear simulation program, MASTA. Lastly, NSGA-II optimization method was used to find the best combination of design parameters. As a result of the optimization, the total weight, transmission error, misalignment, face load distribution factor were all reduced, and the maximum stress was also shown to be a safe level, confirming that the overall gear performance was improved.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.36 no.2
• /
• pp.121-130
• /
• 2023

Finite element analysis is widely used to predict the structural stability and tooth contact performance of gears. This study focused on the effect of finite element modeling conditions of a spur gear on the simulation result and the model simplification. The gear body and teeth, teeth width, configuration of mesh, frictional coefficient, and simulation time interval (gear mesh cycle division) were selected for model simplification for gear analysis. The static transmission error during a single-gear mesh cycle was calculated to represent the performance of the gear, and the elapsed time was measured as a simplification factor. Contact stress distribution was also checked. The differences in maximum transmission error and elapsed time depending on the model simplification methods were analyzed. After all simplification methods were estimated, an optimal combination of the methods was defined, and the result was compared with that of the most detailed modeling methods.

• Journal of Internet Computing and Services
• /
• v.25 no.2
• /
• pp.1-9
• /
• 2024

Although federated learning is designed to be safer than centralized methods in terms of security and privacy, it still has many vulnerabilities. An attacker performing an adversarial attack intentionally manipulates the deep learning model by injecting carefully crafted input data, that is, adversarial examples, into the client's training data to induce misclassification. A common defense strategy against this is so-called adversarial training, which involves preemptively learning the characteristics of adversarial examples into the model. Existing research assumes a scenario where all clients are under adversarial attack, but considering the number of clients in federated learning is very large, this is far from reality. In this paper, we experimentally examine aspects of adversarial training in a scenario where some of the clients are under attack. Through experiments, we found that there is a trade-off relationship in which the classification accuracy for normal samples decreases as the classification accuracy for adversarial examples increases. In order to effectively utilize this trade-off relationship, we present a method to perform adversarial training by adaptively selecting a loss function depending on whether the client is attacked.

• Journal of Internet Computing and Services
• /
• v.19 no.5
• /
• pp.67-75
• /
• 2018

According to Symantec's Internet Security Threat Report(2018), Internet security threats such as Cryptojackings, Ransomwares, and Mobile malwares are rapidly increasing and diversifying. It means that detection of malwares requires not only the detection accuracy but also versatility. In the past, malware detection technology focused on qualitative performance due to the problems such as encryption and obfuscation. However, nowadays, considering the diversity of malware, versatility is required in detecting various malwares. Additionally the optimization is required in terms of computing power for detecting malware. In this paper, we present Stream Order(SO)-CNN and Incremental Coordinate(IC)-CNN, which are malware detection schemes using CNN(Convolutional Neural Network) that effectively detect intelligent and diversified malwares. The proposed methods visualize each malware binary file onto a fixed sized image. The visualized malware binaries are learned through GoogLeNet to form a deep learning model. Our model detects and classifies malwares. The proposed method reveals better performance than the conventional method.

• Journal of Internet Computing and Services
• /
• v.19 no.1
• /
• pp.1-10
• /
• 2018

Network threats such as Internet worms and computer viruses have been significantly increasing. In particular, APTs(Advanced Persistent Threats) and ransomwares become clever and complex. IDSes(Intrusion Detection Systems) have performed a key role as information security solutions during last few decades. To use an IDS effectively, IDS rules must be written properly. An IDS rule includes a key signature and is incorporated into an IDS. If so, the network threat containing the signature can be detected by the IDS while it is passing through the IDS. However, it is challenging to find a key signature for a specific network threat. We first need to analyze a network threat rigorously, and write a proper IDS rule based on the analysis result. If we use a signature that is common to benign and/or normal network traffic, we will observe a lot of false alarms. In this paper, we propose a scheme that analyzes a network threat and extracts key signatures corresponding to the threat. Specifically, our proposed scheme quantifies the degree of correspondence between a network threat and a signature using the LDA(Latent Dirichlet Allocation) algorithm. Obviously, a signature that has significant correspondence to the network threat can be utilized as an IDS rule for detection of the threat.

• Journal of Internet Computing and Services
• /
• v.20 no.6
• /
• pp.11-20
• /
• 2019

Numerous enterprises, organizations and individual users are exposed to large DDoS (Distributed Denial of Service) attacks. DDoS attacks are performed through a BotNet, which is composed of a number of computers infected with a malware, e.g., zombie PCs and a special computer that controls the zombie PCs within a hierarchical chain of a command system. In order to detect a malware, a malware detection software or a vaccine program must identify the malware signature through an in-depth analysis, and these signatures need to be updated in priori. This is time consuming and costly. In this paper, we propose a botnet detection scheme that does not require a periodic signature update using an artificial neural network model. The proposed scheme exploits Word2Vec and accelerated hierarchical density-based clustering. Botnet detection performance of the proposed method was evaluated using the CTU-13 dataset. The experimental result shows that the detection rate is 99.9%, which outperforms the conventional method.