순환신경망을 활용한 콘크리트궤도의 장기 침하 거동 예측

김준영,이수형,최영태,우상인,Kim, Joonyoung,Lee, Su-Hyung,Choi, Yeong-Tae,Woo, Sang Inn 한국지반공학회 2020 한국지반공학회논문집 Vol.36 No.3

콘크리트궤도는 고속, 고밀도의 운행선로에서도 차량의 주행안정성이 높고, 이용승객에게 좋은 승차감을 제공할 수 있으며, 궤도 보수 비용을 대폭 절감시킬 수 있는 장점을 가지고 있어 고속선을 중심으로 점차 적용 구간이 늘어나고 있다. 하지만, ◯◯고속철도 토공 구간의 노반 침하로 기인된 콘크리트궤도의 2차적 침하가 발생하는 사례가 빈번히 발생하고 있어, 이로 인한 철도 운행의 안정성 문제가 대두되고 있다. 허용 잔류침하량을 초과하는 콘크리트궤도에 대한 효율적인 유지보수 방안 수립을 위해서는 공용 기간 중 발생하는 콘크리트궤도의 장기 침하 거동을 합리적으로 예측하는 것이 필수적이다. 따라서, 본 연구에서는 다양한 인공지능 기술 중 순환신경망을 활용하여 콘크리트궤도의 장기 침하 거동을 예측하는 모델을 개발하였다. 또한 기존 모델의 침하 예측 결과와 비교를 통해 개발된 모델의 적용 가능성을 모색하였다. The railway concrete track has been increasingly adopted for high-speed train such as KTX due to its high running stability, improved ride quality for the passengers, and low maintenance cost. However, excessive settlement of the railway concrete track has been monitored at embankment sections of the ◯◯ High-speed Line, resulting in the concerns on the safety of railway operation. In order to establish an effective maintenance plan for the concrete track railway exceeding the allowable residual settlement, it is essential to reasonably predict their long-term settlement behavior during the public period. In this study, we developed a model for predicting the long-term settlement behavior of concrete track using recurrent neural network (RNN) and examined the applicability of the developed model.

Is Good Beginning Half Done?

JoonYoung Kim(김준영) 한국노동연구원 2008 노동정책연구 Vol.8 No.1

일본 일반사회조사(JGSS)의 데이터를 이용해서, 본 연구는 일본에서의 첫 일자리의 고용형태가 이후의 고용전망에 미치는 효과를 분석하였다. 첫 일자리 고용 지위의 잠재적 내생성을 통제하기 위해서, 본 연구는 단일변량 프로빗 모델과 이변량 프로빗 모델을 함께 사용하여 분석하였으나. 첫 일자리 고용 지위의 내생성에 대한 증거는 발견되지 않았다. 최종 학교를 마친 후 비정규직으로 첫 취업을 한 노동자는 정규직으로 첫 취업한 노동자에 비해서 현재 정규 일자리를 가질 확률이 유의하게 낮았다. 또한 본 연구는 비정규직 첫 취업 효과의 민감도(sensitivity)를 평가하기 위해 여러 모형을 설정하여 분석하였으나, 비정규직 첫 취업의 부정적 효과는 대체로 안정적이었다. Using data from Japanese General Social Surveys (JGSS), this study examines the effects of non-regular initial employment status on future employment prospects. In order to deal with the potential endogeneity of initial employment status, empirical analyses use bivariate probit models as well as univariate probit models. However I found almost no evidence that the univariate probit models were subject to endogeneity bias. The empirical results indicate that getting non-regular jobs at the year of leaving school reduces significantly the probability of obtaining regular full-time job at present. I also implemented several sensitivity tests, and the finding were relatively stable across several specifications.

비선형 관측기를 이용한 차량의 타이어 횡력 감지시스템 개발

김준영(Joonyoung Kim),허건수(Kunsoo Huh) 한국자동차공학회 2000 한국 자동차공학회논문집 Vol.8 No.4

Longitudinal and lateral forces acting on tires are known to be closely related to the tract-ability, braking characteristics, handling stability and maneuverability of ground vehicles. In this paper, in order to develop tire force monitoring systems, a monitoring model is proposed utilizing not only the vehicle dynamics but also the roll motion. Based on the monitoring model, three monitoring systems are developed to estimate the tire force acting on each tire. Two monitoring systems are designed utilizing the conventional estimation techniques such as SMO (Sliding Mode Observer) and EKF (Extended Kalman Filter). An additional monitoring system is designed based on a new SKFMEC (Scaled Kalman Filter with Model Error Compensator) technique which is developed to improve the performance of EKF method. Tire force estimation performance of the three monitoring systems is compared in the Matlab simulations where true tire force data is generated from a 14 DOF vehicle model with the combined-slip Magic Formula tire model. The real-time estimation performance is also evaluated in the HILS steering simulator which was built in our Lab. Simulation results show that the SKFMEC method gives the best performance when the driving and road conditions are perturbed.

[전기·전자] 비선형 관측기를 이용한 차량의 타이어 횡력 감지시스템 개발

김준영(Joonyoung Kim),허건수(Kunsoo Huh) 한국자동차공학회 1999 한국자동차공학회 춘 추계 학술대회 논문집 Vol.1999 No.11_2

longitudinal and lateral forces acting at tires are known to be closely related to the tractive ability, braking characteristics, handling stability and maneuverability of ground vehicles, However, it is not feasible in the operating vehicles to measure the tire forces directly because of high cost of sensors, limitations in sensor technology, imerference with the tire rotation and harsh environment. Another approach is the indirect sensing technique which estimates the tire forces based on the remote sensor outputs and vehicle dynamics models, In this paper, in order to develop tire force monitoring systems, a new vehicle dynamics monitoring model is proposed including the roll motion, Based on the monitoring model, three different tire force monitoring systems are designed to estimate the lateral tire force acting at each tire. Two monitoring systems are developed utilizing the ennvernional estimation methods such as EKF (Extended Kalman Filter) and SMO (Sliding Mode Observer), In addition, a newly proposed SKFMEC (Scaled Kalman Filter with Model Error Compensator) method is developed utilizing the EKF method. Tire force estimation performance of the three monitoring systems are compared in the Matlab simulations where true tire force data is generated from a 14 DOF vehicle model with a combined-slip Magic Formula tire model. The real-time estimation performance is also evaluated in the HiLS steering simulator which was built in the lab. for this purpose

손상 하수관으로 인한 지반함몰의 위험도 평가를 위한 랜덤 포레스트 모델 개발

김준영(Kim, Joonyoung),강재모(Kang, Jae Mo),백성하(Baek, Sung-Ha) 한국지반공학회 2021 한국지반공학회논문집 Vol.37 No.12

Efficient Crowdsourcing System for Image Object Localization

Joonyoung Kim(김준영),Donghyeon Lee(이동현),Kyomin Jung(정교민) 한국정보과학회 2018 한국정보과학회 학술발표논문집 Vol.2018 No.12

원심모형실험을 통한 원형 수직구 굴착 중 발생하는 지반 변형 평가

김준영(Kim, Joonyoung) 한국지반공학회 2022 한국지반공학회논문집 Vol.38 No.1

텍스트 마이닝과 소셜네트워크분석을 통한 AI·빅데이터 기반 제조기술 트렌드 연구

김일중(Iljung Kim),한아람(Aram Han),채희수(Heesu Chae),김하정(Hajeong Kim),유승화(Seunghwa Ryu),김준영(Joonyoung Kim),신민수(Minsoo Shin),김흥남(Heungnam Kim) 한국기술혁신학회 2021 기술혁신학회지 Vol.24 No.5

최근 스마트공장 질적 고도화의 일환으로 제조 생태계에 AI·빅데이터 기술적용에 대한 관심이 고조되고 있다. 그러나 기존 제조기술 트렌드 연구는 스마트공장 구축을 위한 자동화·로봇화에 집중되어 제조 AI·빅데이터 중심의 제조기술 트렌드 연구 및 활용은 미비한 실정이다. 본 연구는 제조업에 적용되는 AI·빅데이터 기술에 대한 최근 6년 동안의 주요 이슈 및 트렌드를 분석하여 제조기업의 AI·빅데이터 기술도입의 방향성을 제시하고자 한다. 분석결과의 적시성과 정확도를 증대하기 위해 특허, 논문, 뉴스, 박람회, 인공지능 중소벤처 제조플랫폼(KAMP) 콘텐츠를 분석 데이터로 통합화하였다. 그리고 수집 데이터의 정제로 Python 3.6과 텍스트 마이닝 및 소셜네트워크분석은 KrKwic 2.0, UCINET 6.721을 활용하였다. 분석 결과, 시간의 흐름에 따라 제조 특화 AI, 딥러닝, 알고리즘 키워드의 연결중심성이 높아지는 것으로 나타났으며 이를 구현하기 위해 고성능 컴퓨팅 기술을 포함한 다수의 제조AI 기술군집이 형성되었다. 제조현장의 AI 적용목적에 따라 보다 다양하고 최적화된 제조 특화 AI 알고리즘 및 제조 빅데이터 융·복합기술의 중요도가 증가하고 있음을 관찰할 수 있었다. Recently, there has been a surge of interest in manufacturing AI and big data technology application. Since current manufacturing technology trend research is focused on automation and robotization for the construction of smart factories, study and utilization of manufacturing technology trends centered on manufacturing AI and big data are insufficient. Therefore, this study examines major issues and trends in AI·big data technologies applied to the manufacturing industry over the last six years. To reflect the timeliness and accuracy of analysis results, patents, theses, news, fairs, and Korea AI Manufacturing Platform(KAMP) contents are integrated into analysis data, and data refinement is performed using Python 3.6, text mining, and social network analysis. KrKwic 2.0 and UCINET 6.721 were utilized for this purpose. It was discovered that the degree centrality of manufacturing AI keywords rose with time, prompting the formation of a number of manufacturing AI technology clusters. The significance of more diverse AI algorithms specialized in manufacturing and high-performance infra resources increases depending on the aim of AI application in the manufacturing site.

IT 분야의 융합교육에 대한 전문대학 교수자와 학습자의 인식 비교

김경아(Kyong Ah Kim),김지심(Ji Sim Kim),권수현(Suhyun Kwon),김준영(Joonyoung Kim),진명숙(Myung Sook Jin) 한국컴퓨터정보학회 2019 한국컴퓨터정보학회 학술발표논문집 Vol.27 No.2

정화 보조지표와 시료 채취 방법 제안을 통한 토양정화검증 제도 개선 연구

권지철,이군택,김태승,윤정기,김지인,김용훈,김준영,최정민,Kwon, Ji Cheol,Lee, Goontaek,Kim, Tae Seung,Yoon, Jeong-Ki,Kim, Ji-in,Kim, Yonghoon,Kim, Joonyoung,Choi, Jeongmin 한국지하수토양환경학회 2016 지하수토양환경 Vol.21 No.6

In addition to the measurement of the concentration of soil contaminants, the new idea of indicative parameters was proposed to validate the remedial works through the monitoring for the changes of soil characteristics after applying the clean up technologies. The parameters like CFU (colony forming unit), pH and soil texture were recommended as indicative parameters for land farming. In case of soil washing, water content and the particle size distribution of the sludge were recommended as indicative parameters. The sludge is produced through the particle separation process in soil washing and it is usually treated as a waste. The parameters like water content, organic matter content, CEC (cation exchange capacity) and CFU were recommended as indicative parameters for the low temperature thermal desorption method. Besides the indicative parameter, sampling methods in stock pile and the optimal minimum amount of composite soil sample were proposed. The rates of sampling error in regular grid, zigzag, four bearing, random grid methods were 17.3%, 17.6%, 17.2% and 16.5% respectively. The random grid method showed the minimum sampling error among the 4 kinds of sampling methods although the differences in sampling errors were very little. Therefore the random grid method was recommended as an appropriate sampling method in stock pile. It was not possible to propose a value of optimal minimum amount of composite soil sample based on the real analytical data due to the dynamic variation of $CV_{fund{\cdot}error}$. Instead of this, 355 g of soil was recommended for the optimal minimum amount of composite soil sample under the assumption of ISO 10381-8.