http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Data Augmentation Techniques of Power Facilities for Improve Deep Learning Performance
장승민,손승우,김봉석,Jang, Seungmin,Son, Seungwoo,Kim, Bongsuck 한국전력공사 2021 KEPCO Journal on electric power and energy Vol.7 No.2
Diagnostic models are required. Data augmentation is one of the best ways to improve deep learning performance. Traditional augmentation techniques that modify image brightness or spatial information are difficult to achieve great results. To overcome this, a generative adversarial network (GAN) technology that generates virtual data to increase deep learning performance has emerged. GAN can create realistic-looking fake images by competitive learning two networks, a generator that creates fakes and a discriminator that determines whether images are real or fake made by the generator. GAN is being used in computer vision, IT solutions, and medical imaging fields. It is essential to secure additional learning data to advance deep learning-based fault diagnosis solutions in the power industry where facilities are strictly maintained more than other industries. In this paper, we propose a method for generating power facility images using GAN and a strategy for improving performance when only used a small amount of data. Finally, we analyze the performance of the augmented image to see if it could be utilized for the deep learning-based diagnosis system or not.
딥러닝 훈련을 위한 GAN 기반 거짓 영상 분석효과에 대한 연구
장승민,손승우,김봉석,Seungmin, Jang,Seungwoo, Son,Bongsuck, Kim 한국전력공사 2022 KEPCO Journal on electric power and energy Vol.8 No.2
To inspect the power facility faults using artificial intelligence, it need that improve the accuracy of the diagnostic model are required. Data augmentation skill using generative adversarial network (GAN) is one of the best ways to improve deep learning performance. GAN model can create realistic-looking fake images using two competitive learning networks such as discriminator and generator. In this study, we intend to verify the effectiveness of virtual data generation technology by including the fake image of power facility generated through GAN in the deep learning training set. The GAN-based fake image was created for damage of LP insulator, and ResNet based normal and defect classification model was developed to verify the effect. Through this, we analyzed the model accuracy according to the ratio of normal and defective training data.
추적가스를 이용한 의료시설의 감염위험률 현장 평가에 대한 사례연구
장승민(Seungmin Jang),임법규(BeopGyu Lim),조진균(Jinkyun Cho) 대한설비공학회 2023 설비공학 논문집 Vol.35 No.10
This case study analyzed the field evaluation of cross-infection and infection risk using the tracer gas method in mobile infectious disease isolation facilities, focusing on the concept of indoor air dilution through a three-step protocol. Protocol A determines the release concentration of the tracer gas to be used in the experiment, Protocol B presents a method for measuring cross-infection in a single room, and Protocol C involves a statistical analysis of measurement uncertainties to enhance accuracy and reliability. Based on the presented field evaluation, it is possible to assess the infection risk in an indoor air environment of existing or new medical facilities in practical operation. Therefore, it is considered a valuable operational guideline and verification method to minimize cross-infection, implement an effective air conditioning system, and safeguard healthcare workers health.
장승민(Seung Min Jang),박준형(Jun Hyoung Park),양진송(Jin Song Yang),류경수(Kyung Su Ryu),박정수(Jung Soo Park) 한국철도학회 2018 한국철도학회 학술발표대회논문집 Vol.2018 No.5
철도신호시스템은 대규모 인원수송, 고속운행 등의 특성을 가진 철도의 안전운행과 선로이용률을 증대시키는 핵심요소이다. 철도신호시스템은 초기 수신호 방식에서 통표 및 완목식 신호방식을 거쳐 전기를 이용한 연동장치인 ATS, ATC, ATO와 무선통신기술을 기반으로 하는 CBTC로 발전되면서 무인운전이 가능하게 되었다. 또한 철도신호시스템은 열차가 위급한 상황에 직면하지 않도록 지속적인 감시 및 제어를 수행하고 있으며, 높은 신뢰도를 가진 Fail-safe 원칙을 기반으로하고 있다. 본 연구는 국내철도신호의 역사 및 동향, 철도선진국의 신호시스템을 분석하여 신호시스템 발전에 따른 운전방식의 변화와 철도신호의 표준화 및 정책방향을 제시하고자 한다.
가중중앙값 필터를 이용한 에지 방향성 보정 기반 디인터레이싱 기법
장승민(Seung Min Jang),김영철(Young Chul Kim),홍성훈(Sung Hoon Hong) 대한전자공학회 2009 電子工學會論文誌-SP (Signal processing) Vol.46 No.4
본 논문에서는 에지 방향성 보정을 고려한 효율적인 공간적 디인터레이싱 알고리즘을 제안한다. 기존의 에지기반 디인터레이싱 알고리즘들은 다른 공간적 디인터레이싱 알고리즘들에 비해 시각적으로 우수한 결과를 나타내지만 화소단위의 상관도를 이용하여 화소를 보간하기 때문에 잘못된 에지 방향을 찾게 될 경우 영상에 잡음이 발생한다. 제안한 알고리즘은 이러한 단점을 보완하기위해 보간할 화소 주변의 에지 방향성을 검출하고 구해진 정보에 따라 가중중앙값 필터를 적용하여 에지의 방향성을 보정함으로써 정확한 에지의 방향을 찾아낸다. 실험결과 제안된 방법은 기존의 공간적 디인터레이싱 방법들에 비하여 객관적인 PSNR 성능과 주관적 화질도 우수함을 알 수 있다. In this paper, we propose an efficient deinterlacing algorithm which is an edge dependent interpolation based on edge direction refinement. The conventional edge dependent interpolation algorithms have a visually better performance than any other Intra-field deinterlacing algorithms. However they are very sensitive to noise due to the failure of estimating edge direction. In order to exactly detect edge direction, our method detects edge direction of around interpolated pixel and refines the edge direction using weighted median filter. Simulation results have shown the efficacy of the proposed method with significant improvement over the previous methods in terms of the objective PSNR quality as well as the subjective image quality.