Mobile QR Code QR CODE : The Transactions of the Korean Institute of Electrical Engineers

재생에너지 확대에 따른 국내 계통관성 별 적정 예비력 평가 Optimal reserve evaluation for the system inertia in the future Korean power system considering high penetration of RES

https://doi.org/10.5370/KIEE.2020.70.1.001

조재왕(Jae-Wang Cho) ; 강성우(Sung-Woo Kang) ; 박정후(Jeong-hoo Park) ; 임승혁(Seung-Hyuk Im) ; 이병준(Byong-jun Lee)

As the penetration level of renewable energy sources (RESs) increases, the decrease in the system inertia and spinning reserves will be one of the major challenges for the secure operation of power systems. Recently, in order to secure power system frequency within the limit defined in the grid code, accurate analysis on the frequency responsive reserves is getting important. In this study, method of evaluating the minimum level of inertia along with the required reserves to secure system frequency is developed. Furthermore, maximum allowable penetration level of RESs is analyzed in order to secure frequency after a sudden loss of a single generator unit. And with the evaluated minimum inertia for the secure frequency, this study shows the amount of the increased generation and the system frequency response in the time frame after a single generator or two generator units are tripped off. In the developed method, increase in RES penetration is illustrated as a negative load while considering the merit order of conventional generators. A case study is conducted on the future Korean power system of 2024.

수용용량과 계통 손실을 고려한 스마트 인버터의 최적 Volt-VAR 제어 곡선 선정에 관한 연구 A Study on Selecting Optimal Volt-VAR Control Curve of Smart Inverter Considering Hosting Capacity and System Loss

https://doi.org/10.5370/KIEE.2020.70.1.007

강민재(Min-Jae Kang) ; 김지수(Ji-Soo Kim) ; 송진솔(Jin-Sol Song) ; 최호영(Ho-Young Choi) ; 오형진(Hyung-Jin Oh) ; 박우근(Woo-Geun Park) ; 김철환(Chul-Hwan Kim)

The penetration of Renewable Energy Source (RES) has been increasing because RES is drawing attention by the environmental problem in the worldwide. The power flow can be changed from unidirectional to bidirectional because of increasing the penetration of RES. In bidirectional system, the efficient operation of conventional Voltage Regulators (VRs) is difficult. As one of the various methods to reinforce the conventional VR, the Volt-VAR control of the Smart Inverter (SI) is used for supporting the voltage by using the reactive power of RES. In this paper, the Volt-VAR control curve with the modified dead band is analyzed for the factors affecting the Hosting Capacity (HC) and the enhanced method to obtain the maximum HC. The modified Volt-VAR control curve is applied and the power flow is calculated through Open source Distribution System Simulator (OpenDSS). The maximum HC is obtained through Genetic Algorithm (GA) using MATLAB.

BESS의 충?방전 효율 및 전력가격의 변동폭이 비용절감 효과에 미치는 영향 분석 Impact of BESS Charging/Discharging Efficiency and Electricity Price Fluctuation on the Reduction of Electricity Bill

https://doi.org/10.5370/KIEE.2020.70.1.014

오석화(Seok-Hwa Oh) ; 진영규(Young-Gyu Jin)

In recent years, the battery energy storage system (BESS) has been attracting great attention as an alternative to promoting renewable energy sources. Nevertheless, the market for small-scale BESS is still in its infancy particularly due to its high cost. However, the small-scale BESS for homes and buildings is expected to grow in volume as its cost decreases. The customers utilize the BESS mainly for the purpose of cost reduction, and many studies have been conducted. However, the mutual effects of the BESS charging/discharging efficiency and the electricity price fluctuation are not sufficiently considered in the previous researches. Thus, in this study, their impacts on the reduction of electricity bills are analyzed. Specifically, we derive the relationship between BESS efficiency and the price fluctuation for the optimal BESS operation. Additionally, it is shown through the simulations that they can have significant impacts on the number of charging and discharging operations of the BESS and consequently on the reduction of electricity bills. The results may provide customers and policymakers with insights on how to effectively use and disseminate the BESS.

몬테카를로 시뮬레이션 기반 공동주택 전기자동차 충전수요 예측을 고려한 최대부하 시간 변화 분석 An Analysis on the Change of Peak load time Considering the Forecasting of Electric Vehicle Charging Demand for Apartment Based on Monte Carlo Simulation

https://doi.org/10.5370/KIEE.2020.70.1.021

김치연(Chi-Yeon Kim) ; 김채린(Chae-Rin Kim) ; 조수환(Soo-Hwan Cho)

The penetration of Electric Vehicles(EV) has been increasing worldwide recently to reduce greenhouse gases(GHG), and the government plans to supply about 3 million EV until 2030. The EV charging pattern is determined by the ToU(Time-of-Use) tariff, and is mainly charged from 18:00 to 23:00, which is the middle load period. The EV charging demand will further increase the power demand for apartments, which surges after 18:00 after returning home, so it is necessary to accurately forecast the EV charging demand pattern. In this paper, we determine the probability variables of input data such as SoC(State-of-Charge), charging start time, and charging method of EV users to forecast the EV charging demand pattern based on Monte Carlo simulation. In addition, the changes in the power demand patterns of the apartment according to the EV charging demand patterns are estimated using the actual data of the apartment power demand pattern. Furthermore, we estimate changes in the pattern of power demand and changes in the peak load time according to the ratio of EVs per household of apartments nationwide.

수요 반응 및 배터리 에너지 저장 시스템을 통한 Two-Stage 주파수 조정에서의 임계값 최적화 Threshold Optimization in A Two-Stage Frequency Regulation Through Demand Response and Battery Energy Storage System

https://doi.org/10.5370/KIEE.2020.70.1.031

이 마데 두위 달만타라(I Made Dwi Darmantara) ; 최호영(Ho-Young Choi) ; 임승민(Seung-Min Lim) ; 김철환(Chul-Hwan Kim)

The increasing of Distributed Energy Resources (DERs) interconnection such as PV into the microgrid has increase the case of frequency instability. The combination of Demand Response (DR) and Battery Energy Storage System (BESS) can be used as the two-stage frequency regulation to keep the frequency stability in the system. In this paper, a method to select the optimal BESS threshold value in the DR and BESS two-stage frequency regulation is proposed. The IEEE standard 13-bus distribution network was modified as an islanded microgrid and used to perform the simulation. The simulation results shows that the combined DR and BESS with an optimal BESS threshold value can keep the frequency within the maximum allowable frequency nadir value and reduce the energy usage of BESS.

IEC-61850 비동기 정상전압 및 역상전류를 이용한 차동방식의 송전선로 후비 보호 알고리즘에 대한 연구 A Study on Transmission Line Backup Protection based on Differential Algorithm Using IEC-61850 Asynchronous Positive Sequence Voltages and Negative Sequence Currents

https://doi.org/10.5370/KIEE.2020.70.1.038

고창성(Chang-Sung Ko) ; 이남호(Nam-Ho Lee) ; 남순열(Soon-Ryul Nam)

In this paper, a negative-sequence current differential algorithm using asynchronous positive-sequence voltages is proposed to solve the problem that the negative-sequence current differential algorithm using asynchronous negative-sequence voltages may cause malfunction due to the small magnitude of the negative-sequence voltages. This algorithm using asynchronous negative-sequence voltages is an algorithm for solving erroneous operation due to effects of fault resistance and time synchronization problems of current differential algorithm. And this paper explains why positive-sequence voltage was used, and difference in detection of fault direction at relay point of each algorithm. And then, Each algorithm is tested in an IEC-61850 based centralized protection system environment configured using Real Time Digital Simulator and IEDs(Intelligent Electronic Devices), which are implemented with the MMS-EASE Lite library and Evaluation Modules. The test results show that only the proposed algorithm operate successfully, and therefore the proposed algorithm is appropriate for transmission line backup protection.

심층 신경망을 이용한 재생에너지원의 PMU 빅 데이터기반 새로운 계통 현상 판별 A PMU Big Data based New Systematic Phenomenon Identification of RES using Deep Neural Network

https://doi.org/10.5370/KIEE.2020.70.1.045

이경민(Kyung-Min Lee) ; 박철원(Chul-Won Park)

Recently, for real-time monitoring of renewable energy, a wide area power system monitoring and operation technology using PMU has emerged. Through WAMS based on the PMU, time synchronization data for a wide area is acquired and a vast amount of data is accumulated. Therefore, it is a problem to be solved in the future to process this data and deliver highly usable and valuable system status information to system operators. This paper proposes a new systemic phenomenon identification algorithm using big data of PMU installed in RES and a DNN. First, the PMU installed at the RES in the Gangwon region is introduced, and then the data structure collected is explained. Next, by analyzing each system phenomenon from the PMU data, a total of 8 types of system data such as steady state, tap rise, tap fall, feed-in and out, etc. are generated. After conducting supervised learning by constructing learning data for 8 systematic phenomena using a DNN, systematic phenomena discrimination is performed on the DNN model learned through the test data. Finally, the algorithm was designed, implemented, and evaluated to identify robust systematic phenomena for new PMU based big data. The simulation results showed that the proposed new algorithm accurately discriminates all systematic phenomena.

ESS용 연계변압기의 포화에 따른 철공진 특성 모델링에 관한 연구 A Study on Modeling of Ferro-resonance Characteristics with Saturation of Grid-connected Transformer in Energy Storage System

https://doi.org/10.5370/KIEE.2020.70.1.051

태동현(Dong-Hyun Tae) ; 이후동(Hu-Dong Lee) ; 한병길(Byeong-Gill Han) ; 노대석(Dae-Seok Rho)

Battery system might be severely effected by unexpected ferro-resonance phenomenon in Li-ion battery based ESS, which is energetically installed and operated in large scale way. Therefore, this paper defines mechanism of the ferro-resonance phenomenon with L-C series and series-parallel combination circuits, which is regarded as one of electrical hazards, and proposes characteristics of the ferro-resonance through graphical solutions and equivalent circuit. And also, this paper proposes an algorithm to prevent ferro-resonance phenomenon in the interconnected transformer that can avoid the ferro-resonance by calculating appropriate capacity of PCS filter based on the presented ferro-resonance characteristics. Furthermore, this paper performs modeling of ESS based on PSCAD/EMTDC, which is composed of 3-phase power supply system including circuit breaker and interconnected transformer, PCS(power control system) and battery system. From the simulation results based on the proposed modeling, it is confirmed that secondary voltage of interconnected transformer for ESS can be rapidly increased due to the ferro-resonance phenomenon in L-C series and series-parallel combination circuits and then the battery system may be severely affected by such rapid over-voltage. And also, it is confirmed that the proposed prevention algorithm of ferro-resonance can contribute to stable operation of ESS by installing optimal capacity of filter in PCS.

리튬이온전지를 사용하는 ESS에서 CMV 및 누설전류의 발생원인과 저감방법에 관한 연구 A Study on the Cause of CMV and Leakage Current, and Mitigation Strategy on the Lithium-ion Battery of ESS

https://doi.org/10.5370/KIEE.2020.70.1.061

김승호(Seung-Ho Kim) ; 최형석(Hyoung-Seok Choi) ; 김진용(Jin-Yong Kim) ; 한 아(Ah Han)

ESS(Energy Storage Systems) using lithium ion battery are rapidly increasing their capacity because they can efficiently store electrical energy in a small area with high energy density. However, 29 fire accidents have been reported in Korea for last 4 years. Up to now, the root cause of ESS fire has not been clearly defined, but some lithium ion battery manufacturer assert that the reason of fire may be caused by CMV(Common Mode Voltage) and leakage current. This study investigated the mechanism, measuring, acceptable level, and mitigation of CMV and leakage current through experimental field analysis with its corresponding simulation. The stray capacitance of IGBT module and lithium-ion battery are investigated as the cause of CMV and leakage current, and the magnitude variation of CMV and leakage current by the type of system earthing of the AC side is investigated too, then this paper suggested the use of an isolation transformer as a mitigation method. In addition, it was analyzed how the use of an isolation transformer can reduce CMV and this study proved that the installation of an isolation transformer is a complete and fundamental solution to reduce both CMV and leakage current by simulation and field measurement results. On the other hand, this study found the shape and period of the waveform of CMV and leakage current are completely identical, and prove that these two are produced by the same cause and the leakage current was also demonstrated to be lowered when the CMV was lowered. On the other hand, based on the analyzing the path of the leakage current flow in the grounded system and the non-ground systems, this study found that it was reasonable to measure CMV and leakage current on the DC side of PCS, and this was proved by P-Sim simulation and field experiments.

Random Walk와 Simplex 방법을 이용한 V-shape 매입형 영구자석 동기전동기의 코깅토크 저감 최적 설계 Optimal Design of V-Shape IPMSM Using Random Walk and Simplex Method for Reducing Cogging Torque

https://doi.org/10.5370/KIEE.2020.70.1.072

오제민(Je Min Oh) ; 방태경(Tae-Kyoung Bang) ; 신경훈(Kyung-Hun Shin) ; 최장영(Jang-Young Choi) ; 조한욱(Han-Wook Cho)

This paper presents random walk and simplex optimization algorithm on an optimal design of V-shape interior permanent synchronous motor (V-shape IPMSM). The IPMSM can easily generates a cogging torque because of change in magnetic circuit by interaction between permanent magnets and stator slots. Therefore minimization of the cogging torque, which generates vibration and noise, is an important factor in design of IPMSM. This paper proposes the optimal shape design of V-shape IPMSM using random walk and simplex method for minimizing cogging torque.

WBG 전력 반도체를 위한 넓은 전류 도통 영역을 갖는 전력 회로 보드의 집중정수 모델링 기반 파워 루프 해석 기법 Lumped Parameter Modeling based Power Loop Analysis Technique of Power Circuit Board with Wide Conduction Area for WBG Semiconductors

https://doi.org/10.5370/KIEE.2020.70.1.079

조민신(Min-Shin Cho) ; 김래영(Rae-Young Kim)

In this paper, we propose a power loop analysis method based on lumped parameter modeling of power circuit board with wide current conduction area for WBG power semiconductors. The proposed analysis method is modeled with lumped parameter so that power loops having various current paths can be analyzed, so the analysis is simple, easy to apply, and has the advantage of enabling dynamic power loop analysis. The validity of the lumped parameter model was verified through LTSPICE and Q3D simulation results.

친환경 전력 케이블용 절연체에서 h-BN 충진제가 공간전하와 전기전도 특성에 미치는 영향 Effect of h-BN Filler on the Space Charge and Electrical Conduction Properties of Eco-friendly Power Cable Insulators

https://doi.org/10.5370/KIEE.2020.70.1.089

김철호(Chul-Ho Kim) ; 이준호(June-Ho Lee)

As the demand for HVDC increases, interest in the accumulation of space charges under DC electric field is increasing. In addition, interest in eco-friendly insulating materials that can replace XLPE, an existing power cable insulating materials, is steadily increasing. Therefore, in this study, we prepared XLPE, eco-friendly insulating materials and sample in which nano-particle(h-BN: hexagonal boron nitride) were added to eco-friendly insulating material. The space charge distribution and conductivity were measured and analyzed for the prepared samples, and the space charge and conductivity characteristics of each sample were compared. As results, in space charge characteristics, eco-friendly insulating materials showed relatively excellent characteristics compared to XLPE. However, it was observed that the space charge characteristics of sample added with h-BN was not good compared to samples not added. In the conduction characteristics, eco-friendly materials showed excellent characteristics compared to XLPE.

모터의 전류치를 활용한 전동차 출입문의 특징추출 및 고장진단 예지 알고리즘 선정 방안 연구 A study on the Feature extraction of the Rolling Stock door using the current value of the motor and the selection of a failure diagnosis prediction algorithm

https://doi.org/10.5370/KIEE.2020.70.1.096

임종국(Jong-Kook Lim) ; 윤희중(Hee-Jung Yoon)

Recently, the manufacturing requirements of Rolling Stock are required to diagnose and manage the condition of major safety devices (doors, brake, signals, etc.) to suggest a plan for efficient maintenance. Based on these backgrounds and technology trends, research on Condition-Based Maintenance (CBM) and Prognostic Health Management (PHM) has been actively conducted in recent years. In this study, the current value of the engine drive motor of the door of a Rolling Stock is measured by dividing it into four classes (normal open/abnormal open, normally closed/abnormally closed), and statistically analyzed 13 factors of time domain statistics based on time domain statistics. After verifying the significance, suitable Features were extracted. Based on the machine learning theory, a predictive algorithm that can classify the extracted Features was selected, and the accuracy was verified against the actual measured data class with the selected prognosis algorithm.

컴퓨터 보조 진단 시스템 성능 개선을 위한 새로운 이미지 증대 기법 A New Image Augmentation Method for Improving Computer-aided Diagnosis System Performance

https://doi.org/10.5370/KIEE.2020.70.1.102

이신애(Sin-ae Lee) ; 조현종(Hyun-chong Cho)

Gastric cancer is the largest percentage of cancer cases in Korea. A precise way to find the occurrence of gastrointestinal diseases is through gastroscopy by a trained diagnostic physician. Computer-aided diagnosis (CADx) system helps improve the reliability and speed of diagnosis. The CADx system has developed with deep learning, which is data dependent. However, medical image data is labor intensive and time consuming, making it difficult for large data sets to be formed. To solve this problem, it is important to apply augmentation techniques. In this paper, we propose an augmentation method which is suitable for the data. A basic classification model was made by leaning a data set consisting of the original images. Each of the 14 augment technique-applied data set was input into the generated model and the f1-score values were compared. The f1-score of the highest performance among the proposed methods, was 0.9221, with an increase of about 0.085.

음성 특징과 텍스트 임베딩을 이용한 멀티모달 감정인식 Multi-modal Emotion Recognition using Speech Features and Text Embedding

https://doi.org/10.5370/KIEE.2020.70.1.108

김주희(Ju-Hee Kim) ; 이석필(Seok-Pil Lee)

Many studies have been conducted emotion recognition using audio signals as it is easy to collect. However, the accuracy is lower than other methods such as using facial images or video signals. In this paper, we propose an emotion recognition using speech signals and text simultaneously to achieve better performance. For training, we generate 43 feature vectors like mfcc, spectral features and harmonic features from audio data. Also 256 embedding vectors is extracted from text data using pretrained Tacotron encoder. Feature vectors and text embedding vectors are fed into each LSTM layer and fully connected layer which produces a probability distribution over predicted output classes. By combining the average of both results, the data is assigned to one of four emotion categories : anger, happiness, sadness, neutrality. Our proposed model outperforms previous state-of-the-art methods when they use Korean emotional speech dataset.

머신러닝을 이용한 검색량 기반 악성 URL 탐지기법 연구 A Study on the Detection Method of Malicious URLs based on the Internet Search Engines using the Machine Learning

https://doi.org/10.5370/KIEE.2020.70.1.114

김종관(Jongkwan Kim) ; 장민해(Minhae Jang) ; 임선아(Suna Lim) ; 김명수(Myongsoo Kim)

As internet technology develops and there are increases in cyber activities such as e-mail, SNS and e-commerce, cyber security is being emphasized as one of the most important social issues. In particular, spam mail, SMS phishing and other malicious URLs to breach personal data are inflicting financial and physical damage. This paper proposes a malicious URL detection method based on search volume from internet search engines to resolve such problem. The search volume-based malicious URL detection method uses portal search engines to automatically search URLs linked to emails, text messages and SNS to determine whether or not they can be categorized as malicious URLs according to the volume of the search result. This method allows normal URLs, but will protect to abnormal behaviors.

송전선로 드론점검을 위한 딥러닝 기반 자동촬영 카메라짐벌 시스템 개발 Development of Deep Learning-based Automatic Camera Gimbal System for Drone Inspection of Transmission Lines

https://doi.org/10.5370/KIEE.2020.70.1.121

류서현(Seo-Hyeon Ryu) ; 박준영(Joon-Young Park) ; 김석태(Seok-Tae Kim) ; 김태원(Tae-Won Kim) ; 고병성(Byung-Sung Ko) ; 우정욱(Jung-Wook Woo)

In this paper, a new automatic camera gimbal system was developed to efficiently inspect transmission lines using a drone. As drone technology has advanced tremendously over the past few years, drones are being used to replace dangerous tasks in the electric power industry. Especially, Korea Electric Power Corporation has used the automatic drone inspection system developed for power transmission lines, but while this drone system flies on autopilot, its camera gimbal is still controlled manually. Moreover, the camera gimbal control in the field was often interrupted by electromagnetic interference from ultra-high voltage power lines. To overcome this problem, we developed the new camera gimbal system that can automatically shoot power facilities on the basis of Deep Learning. Its control algorithms mainly consists of a photographing algorithm for a steel tower, a photographing algorithm for power conductors, and an automatic drone landing algorithm. This fully automated system is expected to greatly increase the task efficiency and accuracy for transmission line inspection without any kinds control efforts.

Off-Policy 학습 기반 LQR을 이용한 파라미터 추정 및 전력 시스템 동기 탈조 판별을 위한 외란 관측기 설계 Parameter Estimation Using an Off-Policy Learning-Based LQR and Its Application to a Disturbance Observer Design for Out of Step Detection of a Power System

https://doi.org/10.5370/KIEE.2020.70.1.130

양선직(Sun Jick Yang) ; 장수영(Su Young Jang) ; 손영익(Young Ik Son)

In this paper, a reinforcement learning-based Linear Quadratic Regulator(LQR) design method has been adopted to identify unknown system parameters. The off-policy learning-based LQR can obtain the optimal control gain through an iteration technique known as policy iteration, without using the system model parameters. Augmented states, using the system output integration, can help to alleviate the rank condition on the proposed parameter estimation method. Increasing the system model information accuracy allows the disturbance observers to detect various system faults with the least amount of estimation error. The line fault detection ability of a power system for out-of-step prediction has been studied by applying the proposed parameter estimation scheme to a single machine infinite bus system. Simulation results show that both the system parameters and the external disturbance can be successfully estimated through the proposed method

CNN을 이용한 세포영상 자동분류 알고리즘에 관한 연구 Automatic cell image classification with convolutional neural networks

https://doi.org/10.5370/KIEE.2020.70.1.139

김상희(Sang-Hee Kim) ; 이재형(Jae-hyung Lee) ; 최은영(Eun-Young Choi) ; 전성태(Sung-tae Jeon) ; 최민영(Min-young Choi) ; 조서현(Seo-hyun Jo) ; 최세운(Se-woon Choe)

Recently, artificial intelligence can be used in various fields, especially for medical purposes. For example, it can help diagnose lung diseases and cancer accurately and quickly, thereby reducing the time and cost of medical treatment. In this study, image data were acquired using cultured cervical cancer cells and skin fibroblast cells. The acquired images were pre-processed using OpenCV and enabled the creation of input data optimized for training. In addition, an optimal deep learning algorithm was designed to classify cells by type using transfer learning methods. As a result, the CNN-based learning and automatic classification method proposed in this experiment showed a high accuracy of over 98% and is expected to be used for accurate diagnosis and treatment of diseases in the future.

다관절 로봇을 위한 TD3-GAIL 기반의 가상 모방 학습 방법 Virtual Imitation Learning method based on TD3-GAIL for robot manipulator

https://doi.org/10.5370/KIEE.2020.70.1.145

조성현(Seonghyeon Jo) ; 박종천(Jongcheon Park) ; 이상문(Sangmoon Lee)

As robots replace human work, interests in robot teaching are increasing to implement various tasks and to quickly generate task motions. However, there are some difficulties to apply into articulated robots with multiple degrees of freedom. To solve this problem, we developed a virtual environment of robot manipulator for imitation learning and proposed a twin delayed deterministic generative adversarial imitation learning method which provides the imitation of expert’s demonstration. The developed virtual environment provides an efficient way to gather expert’s demonstration data. The learning performance of the proposed method was verified by comparing it with the conventional one based on the virtual environment for robot manipulators.

딥러닝 기반 객체 탐지 알고리즘을 이용한 대장내시경 용종 탐지 시스템 Automated Polyp Detection System in Colonoscopy using Object Detection Algorithm based on Deep Learning

https://doi.org/10.5370/KIEE.2020.70.1.152

이정남(Jeong-nam Lee) ; 조현종(Hyun-chong Cho)

In Korea, colon cancer is increasing due to westernized eating habits. Colonoscopy is being used to reduce deaths from colon cancer and studies of CADx(Computer-aided Diagnosis) are being developed to improve accuracy. Due to the nature of medical data, it was difficult to collect a lot of data, so data was increased 25 times using AutoAugment’s CIFAR-10 policy, and YOLOv4(You Only Look Once), a real-time object detection algorithm, was used to detect lesions. A new object detection algorithm, YOLOv4, use new eight features such as Weighted-Residual-Connections, Cross-Stage-Partial-connections, Cross mini-Batch Normalization and SelfAdversarial-Training. The performance of augmented data had a maximum mAP of 27.44 higher than the original data. The average IoU(Intersection over Union) was 11.44 higher than the original data. When the IoU value is 0.5, the F1-scores of the original data and the augmented data are 0.9 and 0.97 respectively.

DQN 및 DDQN을 이용한 트레이딩 봇 구현 및 성능 비교 Trading Bot Implementation and Performance Comparison Using DQN and DDQN

https://doi.org/10.5370/KIEE.2020.70.1.158

김민태(Min Tae Kim) ; 김병욱(Byung Wook Kim)

This study addresses a reinforcement learning(RL)-based trading bot that can automatically trade stocks using stock trading data and various indicators. To construct the training data, we used daily basic stock price and technical auxiliary indicator, and input them as state variables to conduct trading actions. By comparing the holding assets and stock prices, a Buy agent that predicts the value of buy/hold behavior or a Sell agent that predicts the value of sell/hold behavior is selected for trading action. For the performance analysis, the returns according to the stock price trend during the training and test period are shown for Deep Q-Network (DQN) and Double DQN (DDQN). Experiments showed that both algorithms for test data showed an increase in assets over the long term and follows the trend of the stock price. While DQN reacts sensitively to changes in stock prices, DDQN reacts relatively monotonically, and thus DDQN can guarantee the increase in assets and stable trading.

딥 러닝을 이용한 새로운 다중 객체 구별 방법 A Novel Multi-Object Distinction Method using deep learning

https://doi.org/10.5370/KIEE.2020.70.1.168

오세운(Se-woon Oh) ; 이창현(Chang-hyun Lee) ; 김선목(Sun-mok Kim) ; 임덕진(Deok-jin Lim) ; 이기백(Ki-beak Lee)

In this paper, we propose a novel multi-object distinction method with class-agnostic object detection and class retreival. Multi-object distinction is usually divided into the processes of detecting and classifying an object. Since it is common for industrial applications to add new kinds of objects to be recognized, it is inefficient to re-train the system every time the new object is added. Thus, the propose method employs two deep learning models to solve this problem. 1) Class agnostic object detection model to predict the bounding boxes regardless of the classes of objects and 2) Class retrieval model to determine the classes of the objects. The experimental results show that the proposed method successfully detects and classifies the both experienced and inexperienced objects: the final classification accuracy for 15 learned objects was 98.0%, and for the other 30 new objects that had not been learned. the accuracy was 87.7% on average.

깊은 합성곱 신경망 모델에 따른 유방 초음파 영상 분류 성능 비교 Comparison of deep convolutional neural networks for classification of breast ultrasound images

https://doi.org/10.5370/KIEE.2020.70.1.176

박주영(Juyoung Park) ; 김이삭(Yisak Kim) ; 유창완(Chang-Wan Ryu) ; 김형석(Hyungsuk Kim)

Breast ultrasound has been widely utilized for classifying tumors into benignancy and malignancy. The limitations of traditional breast ultrasound are the handcrafted features obtained by well-trained sonographers and subjective decision according to different individual experiences. Recently, CNN-based deep learning techniques have exhibited better performance in medical images. However, most research for deep learning in medical ultrasound adopts CNN models developed for natural images due to the lack of common standard and dataset. In this paper, we compare six DCNN models which exhibit good performance for natural images - VGGNet, ResNet, InceptionNet, DenseNet, and EfficientNet. Our classification results demonstrate that CNN models of relatively lower performance on natural images show better performance on gray-scale ultrasound images and further study of CNN models are needed focusing on the features of medical images.

이미지 증대 기법을 이용한 토마토 병충해 분류 Tomato Leaf Disease Classification with Image Augmentation Methods

https://doi.org/10.5370/KIEE.2020.70.1.184

함현식(Hyun-sik Ham) ; 조현종(Hyun-chong Cho)

The tomato is one of important crops in the world market with high commercial value. The early detection of disease is crucial for an successful crop yield. Many studies have recently been conducted to identify plant disease. In this paper, tomato disease classification using leaf images is proposed. Using the convolutional neural network(CNN), the features of disease are extracted and learned to classify. Data augmentation methods, Google’s AutoAugment algorithm and GAN(Generative Adversarial Networks), are used to increase tomato disease data. The classification model classifies nine classes of tomato disease. We compared the original model with the data augmentation models and explored that the classification that produced good performance. As a result, the SVHN policy of AutoAugment model achieved F1 Score 0.945

RVC 정규화와 전이학습을 이용한 손동작 인식 Hand Gesture Recognition using RVC Normalization and Transfer Learning

https://doi.org/10.5370/KIEE.2020.70.1.190

김수열(Su-Yeol Kim) ; 김익진(Ik-Jin Kim) ; 이용찬(Yong-Chan Lee) ; 이연정(Yun-Jung Lee)

In this paper, we propose a new hand gesture recognition strategy using network-based transfer learning(TL) and reference voluntary contraction(RVC) normalization. The structure and parameters of the state-of-the-art deep learning models such as VGG19, ResNet152 and DenseNet121 for source task of image classification are reused in the target task of hand gesture recognition based on surface electromyography(EMG) signals. To mitigate the difficulty in handling the subject-dependent EMG signals, the RVC normalization is adopted in the signal pre-processing. The time-domain EMG signals are transformed into 2-D images for TL networks. The experimental results verify the validity of the proposed method in terms of recognition accuracy. The TL using VGG19, RVC normalization and gray image transformation shows 99.78% accuracy for the data from 15 participants performing 20 different gestures.

YOLO와 CRNN을 사용한 전기 계량기 이미지에서의 카운터 숫자 인식 Counter Digit Recognition in Power Meter Image Using YOLO and CRNN

https://doi.org/10.5370/KIEE.2020.70.1.201

차영화(Young Hwa Cha) ; 박병준(Byung Joon Park)

This study proposes a method for automatically recognizing digits in power meter images. Our targets are analog power meters that display usage with four counters. We employed object detection to find the numeric domain in the meter image. And optical character recognition was performed to recognize digits in the detected numeric area. Two types of deep neural network models are used for object detection and optical character recognition. To train the model and evaluate its performance, we generated datasets from meter images. In this paper, we propose a model in which two types of deep neural networks are connected, and this model has more than 98% recognition performance

서버관리 시스템을 위한 명령형 음성 사용자 인터페이스 설계 및 구현 Design and Implementation of Command-and-Control Voice User Interface for Server Management Systems

https://doi.org/10.5370/KIEE.2020.70.1.207

구흥서(Heung-Seo Koo)

Voice interface technology is a technology that enables dialogue between machines and humans by using ‘words’, one of the most natural means of communication for humans. One of the advantages of using the voice interface when entering commands is that it can be used in parallel with the keyboard interface. Therefore, usability can be improved by supporting a voice user interface in softwares running on computer servers or PCs. In this study, a command-and-control interface was designed and implemented for a multi-server management system that manages multiple servers. The command input mode used voice and keyboard, and the information output mode used the screen and beep interface. The voice recognizing software can be run in a web browser and uses the open source Annyang library. The results of this study are expected to be utilized in the design of voice user interfaces for application software, facility management software, and IoT device monitoring systems for professional users.

변위차 두둑 추종 방식을 활용한 자주식 감자수확기용 조향 제어 시스템 개발 Development of Steering Control System for a Self-Propelled Potato Harvester Using a Displacement Difference Ridge Tracking Method

https://doi.org/10.5370/KIEE.2020.70.1.213

임훈(Hoon Lim) ; 서명국(Myoung Kook Seo) ; 신희영(Hee Young Shin) ; 윤병운(Byong Un Yun) ; 방병주(Byoung Ju Bang) ; 주영훈(Young Hoon Joo)

This paper attempts to introduce the development contents of a steering control system using the displacement difference ridge tracking method of a self-propelled potato harvester. Currently, the self-propelled potato harvester is operated manually, and it is easy to automate because it performs repetitive and regular movements and operations in a regularized ridge area. Therefore, we studied a method that can control the steering so that it does not deviate from the path or invade the side ridge while following the center of the ridge being worked. As a result, the equation was established by Ackerman Steering Kinematics, developed a ridge detection sensor and algorithm, and the optimum steering gain value was obtained through an experiment to verify the performance. The purpose of this study is to increase the work efficiency and convenience of the self-propelled potato harvester used in field farming

차량용 2D 라이다를 이용한 특징정보 기반 확장표적추적 기법 Feature Based Extended Target Tracking Using Automotive 2D LIDAR

https://doi.org/10.5370/KIEE.2020.70.1.224

함다혜(Dahye Ham) ; 조형찬(Hyung-Chan Cho) ; 윤유정(Yoo-Jung Yoon) ; 나원상(Won-Sang Ra) ; 한슬기(Seul-Ki Han)

This paper deals with the design problem of an extended target tracking filter using automotive 2D LIDAR. As a practical way to reduce the dimensionality of the LIDAR point cloud data and to extract an important target feature from it, the Hough transform is applied and the corresponding measurement equation is modeled. It is well-known that the pattern of the LIDAR point cloud varies with the relative geometry due to occlusion, which may leads the severe performance degradation in target tracking. To cope with this problem, a multiple model filter is designed by considering the measurement acquisition hypotheses on the target feature. Through the experiments in real-driving condition, the superior performance of the proposed filter over the existing method is demonstrated.

버그 알고리즘을 이용한 청진음의 음향학적 분석법에 관한 연구 A Study on Acoustic Analysis of Stethoscope Signal using the Burg Algorithm

https://doi.org/10.5370/KIEE.2020.70.1.236

김동준(Dong-Jun Kim)

This study describes an acoustic analysis method of stethoscope signal using the Burg algorithm which is a kind of linear prediction, considering the heartsound production mechanism is linear and can be modelled by an inverse filter of the all-pole model. For this study, normal and heart diseased stethoscope sounds were collected in the pediatrics of an university hospital. The collected signals were preprocessed by noise filtering and analyzed by the Burg algorithm, a kind of linear prediction. The AR spectra and the formants for the heartsounds for the normal and the diseased children are estimated and the results are compared. The spectra showed outstanding differences in morphology and formant frequencies between the normal and the diseased children. Normal children showed relatively low frequency of F1(the first formant) and small negative slope from F1. VSD children revealed stiff slope change around F1 to F3. Spectra of ASD children is similar with the normal case, but have negative values of F3. F1-F2 difference of the functional murmur children were relatively large.

발목 PPG 신호 처리 개선을 위한 Zero Phase Filter 설계 및 시스템 연구 Zero Phase Filter Design and System Study for Improved Handling of ankle PPG Signals

https://doi.org/10.5370/KIEE.2020.70.1.243

이재원(Jae-Won Lee) ; 윤서진(Seo-Jin Yoon) ; 김경호(Kyung-Ho Kim)

One of the areas of steady interest in modern society is health. In addition, in order to keep the individual’s health at its best, the healthcare device market is undergoing many trials and developments. The latest threat to COVID-19 is a major threat to heart disease patients, and it is time for continuous monitoring technology to be needed to prevent it. In particular, PPG sensor technology is relatively commercialized amid various attempts to analyze cardiovascular conditions. In general, we prefer to measure signals from fingers and wrists, which have a stable advantage in obtaining signals. However, there are cases where it is not suitable for the user’s lifestyle, which creates inconvenience. In this study, signal is measured from the ankles to minimize the inconvenience that users may experience when wearing them. And IIR Filter is applied for stable signal acquisition and analysis. Phase Delay, one of the disadvantages of this, is an unnecessary factor in real-time analysis. Zero Phase Filter is designed and applied to the system for further signal analysis. The results of the study compared three types of raw data, IIR Filter data, and Zero Phase Filter data acquired from PPG sensors, and were able to confirm the tendency of Zero Phase Filter data to match the peak points of raw data relative to IIR Filter data. The above results indicate the tendency for in-depth analysis of signals free from Phase Delay in the real-time acquisition of ankle PPG signals. Through peak-to-peak analysis and deep development of algorithms, various applications are expected to be possible by designing stress estimation and fatigue estimation systems that do not constrain users’ upper bodies.

접지봉 매설을 위한 유도공 굴착장치 및 탄소접지동봉 활용 접지저항 경감 방법 Guided hole excavating device for ground Rod construction and the method for reducing ground resistance with Carbon tube ground copper rod

https://doi.org/10.5370/KIEE.2020.70.1.249

김영배(Young-Bae Kim) ; 이기종(Ki-Jong Lee) ; 문현우(Hyun-Woo Moon) ; 손준상(Jun-Sang Son)

In korea, the electric facilities and communication equipments have been built with a grounding construction in order to protect human life from electrical shock of lightning/ surge voltage and secure the performance of electric appliances and rapid emit an impulsive current into underground for safety. Also, the value of grounding resistance of the electric facilities was legislated to certain resistance value in korea and it should be perform the reinforcement construction for maintaining less than the specified value. But it’s very hard to carry out the grounding reinforcement construction without any damage on the seven underground utilities scattered around underground of downtown area. Futhermore, although it is inevitable to use additional resistance reducing materials for qualifying the value, but the porosity of ground and resultant loss of reducer make that uncontrollable. Therefore we had to seek the process of minimizing the economic loss and basically preventing the national property loss, negligent accident, futhermore, appling the method of cost reduction as well.

배전 가공전선 점퍼용 스리브 최적 압축강도 결정에 관한 연구 A study on the determination of the optimal compression strength for overhead distribution line jumper sleeve

https://doi.org/10.5370/KIEE.2020.70.1.253

전시식(Si-Shik Jeon) ; 박철배(Chul-Bae Park) ; 김영달(Young-Dal Kim)

This is a study on the determination of optimal compression strength of jumper sleeves for distribution overhead wires using aluminum conductor. In order to determine the optimal compression strength, problems of the existing oiled compression standards were reviewed, and conductor resistance, porosity, tensile strength tests and statistical analysis were conducted to propose new standards. As a result of the test, it was found that 10[tonf], which is less than the current 13[tonf] compressive strength applied to aluminum conductors, is the optimal compression strength. Therefore, in this paper, We propose to reduce the standard for applying the compressive strength of jumpers from 13[tonf] to 10[tonf] or more, which is a place where tension does not occur like jumper sleeve.