도시 철도에서 다양한 운행 패턴 생성을 위한 스케줄 작성 최적화 모형 연구

박범환 한국철도학회 2020 한국철도학회 학술발표대회논문집 Vol.2020 No.7

미시적 열차 운행 시뮬레이션을 이용한 열차 리스케줄링을 위한 행동 복제 학습

박범환 한국철도학회 2023 한국철도학회논문집 Vol.26 No.3

As the railway network becomes more complex, it is necessary to convert the current train traffic control system, which is based on the experience of the controller, into an intelligent one by using optimization and deep learning to anticipate conflicts between trains more quickly and resolve the conflicts in real time. The purpose of this study is to develop a behavior cloning model that can imitate controller decision-making according to train operation situation using deep learning technology. However, in the current train traffic control system, it is not easy to secure training data that combines train operational information and controller decisions. In this study, we present a method for generating trainig data using OpenTrack simulation, present a deep learning model for behavior cloning, and verify the accuracy of the models.

철도분야에서의 최적화 기법 활용 사례

박범환 대한산업공학회 2006 ie 매거진 Vol.13 No.1

역내 열차 경합 검지 및 해소를 위한 수리 모형 및 해법

박범환,김경민,홍순흠,김영훈 한국철도학회 2007 한국철도학회논문집 Vol.10 No.2

In large station with high density traffic, trains can be hardly controlled by CTC but by station dispatcher because CTC has difficulties in monitoring all states of affairs happening within each station such as departures, arrivals of many trains from different lines and shunting of trains to move between yards and platforms, etc. Therefore the station's dispatcher has to make quick decision about how to reschedule the times and routes for all the trains scheduled within a specific time window. And such decision becomes harder when an unexpected delay occurs because a delay occurring in a train propagates other trains as time goes on. Generally, it is called the conflict detection and resolution to adjust beforehand the distorted schedule due to a delay to original schedule. Our research is different from the state of the arts in that ours determines simultaneously the routes and the times of arrival and departure of trains, although others do only the arrival and departure time of the trains without considering the alternative routes and shunting of the station. This study suggests a mathematical approach for how to detect in advance and resolve efficiently the conflicts occurring within a station and it will be shown how to reduce delay using our approach by means of analysing the schedule of ChyungRyangRi station.

열차운행시뮬레이션을 활용한 열차 리스케줄링을 위한 Imitation Learning에 관한 연구

박범환,서주희,오시은,이용선 한국철도학회 2022 한국철도학회 학술발표대회논문집 Vol.2022 No.10

Batch-Constrained Deep Q-learning을 이용한 열차 리스케줄링 강화 학습 알고리듬 개발

박범환 한국철도학회 2022 한국철도학회 학술발표대회논문집 Vol.2022 No.10

OpenTrack 시뮬레이션을 이용한 현실적 운전선도 작성에 관한 연구

박범환,임나연 한국도시철도학회 2019 한국도시철도학회논문집 Vol.7 No.2

최근 급행화 및 신규 운행 패턴 도입 등 도시철도 운영 상황 변화에 대응하여, 새로운 열차운행 계획 작성 필요성이 증대하고 있다. 이러한 열차 운행 계획에 있어 가장 기초적인 데이터는 역간 운전 시분이다. 역간 운전 시분은 운행하고자 하는 차량의 성능과 그 차량이 지나게 되는 궤도 특성에 따라 거리별 운행속도를 산출함으로써 구할 수 있는데, 이러한 거리별 열차 운행 속도를 도시한 2차원 그래프를 운전선도라 한다. 이러한 운전선도는 대부분 운행 초기에 작성한 것으로, 부본선 설치에 의한 급행 열차의 도입과 같이 새로운 구간이 포함되거나 skip-stop에 의한 운행 패턴이 수정될 경우, 새로운 운전선도 및 신규 구간에 대한 운전시분을 작성해야 한다. 최근 한국에서도 OpenTrack과 같은 열차운행 시뮬레이션프로그램이 많이 언급되고 있지만, 이 프로그램을 이용하여 TPS를 수행할 경우, 운영사가 보유하고 있는 기존의 운전선도와 상당히 차이가 발생함을 확인할 수 있다. 이는 운전선도 작성 알고리즘 및 파라미터가 프로그램 마다 다르기 때문인데, 본 연구는 기존 운전 선도를 최대한으로 모사하기 위한 열차운행시뮬레이션의 활용 방안을 제시할 것이다. 이를 통해 운영 상황 변화에 따른 신규 구간에 대한 운전 선도를 작성하고, 동일한 구간에 대해서는 기존 운전선도와 최대한 유사한 운전선도를 작성할 수 있을 것으로 기대된다. In response to changes in the operating conditions of urban railways, such as the introduction of skip-stop patterns, the necessity of creating a new train operation plan is increasing. The most basic data for the train operation plan is the running time between stations. The inter-station running time can be obtained by calculating the traveling speed according to the performance of the vehicle to be operated and the track characteristics. The two-dimensional graph showing the traveling speed of the train by distance is referred to as the run curve or speed profile. Such run curve was created at the beginning of the operation. However, if the operating conditions are drastically modified, such as the introduction of the express train by installing the secondary line, a new run curve and running times must be developed. Recently, in Korea, train operation simulation programs such as OpenTrack have been widely discussed. However, when the TPS is executed by using this program, it can be confirmed that there is a considerable difference from the existing run curve possessed by the operator. This study suggests how to use the train operation simulation program to simulate the existing run curve as much as possible. This makes it possible to create a run curve for a new section due to changes in the operational situation and to create an run curve as close as possible for the existing sections.

열차운행시뮬레이션을 이용한 혼잡역에서의 반복 회송 스케줄 작성

박범환,이현균 한국도시철도학회 2020 한국도시철도학회논문집 Vol.8 No.4

With the rapidization and direct connection with new line in metropolitan railway lines, train operation plans are gradually becoming more complicated and then there is a growing need to establish a systematic approach based on optimization and simulation. to develop train operation plan. The purpose of this study is to present a method of utilizing train operation simulation to create a train repetition schedule in congested stations to smoothly operate the train schedule given a train schedule. In this study, using the software OpenTrack, we show how to create train repetition schedule at Oido station, where many trains originate and terminate at Suin-Bundang line. 최근 도시 및 광역 철도 노선에서의 급행화 및 신규 노선과의 직결 운행에 따라, 열차운행계획이 점차 복잡해지고 있어, 기존의 수작업에 기초한 열차 운행 계획 작성 및 분석을 최적화 및 컴퓨터 시뮬레이션 프로그램을 활용한 방법론으로 대체할 필요성이 커지고 있다. 본 연구는 열차 스케줄이 주어졌을 때, 그 열차 스케줄을 원활하게 운영하기 위한 혼잡역에서의 열차 반복 스케줄을 작성하기 위한 열차운행 시물레이션의 활용방법을 제시하는 데 목적이 있다. 본 연구에서 활용하는 열차운행시뮬레이션 프로그램은 OpenTrack이며 분석 대상 노선은 최근 개통한 수인분당선 구간 중 오이도~한대앞 구간이며, 오이도역에서의 시종착열차들의 반복 스케줄 및 출도착선 확정을 위한 시뮬레이션 활용방안을 제시하고자 한다.

다양한 정차 패턴을 고려한 열차 노선계획의 수립

박범환,오석문,홍순흠,문대섭 대한교통학회 2005 대한교통학회지 Vol.23 No.6

The line planning problem is to determine the origin and destination stations of the lines with their frequencies so as to meet the OD demands. Since the advent of high speed trains, Korea railway is confronted with the urgent difficulty to reconstruct the line configuration with the frequencies of each line and each fleet type so the demands could be newly created as well as satisfied. Furthermore, the existing trains except the high speed trains suffer from a longer traveling time than before. Now, to reduce the passenger traveling time, the trains with the various halting patterns are run in the same line. Therefore, it is necessary to develop a new line planning model to consider the various halting patterns. Most of studies find the frequencies of each lines which meet the link traffic loads or minimum link frequencies. But these are based on the assumption of all stop patterns. Furthermore, it is not easy to include the actual constraints as like the minimum number of stops at a station, the maximum number of stops of a train, etc. We develop the line planning model considering not only the various halting patterns but also the actual constraints which is based on the multicommodity network flow model with the additional constraints.

미시적 열차 운행 시뮬레이션을 이용한 열차 지연을 고려한 열차 반복 계획 평가

박범환 한국철도학회 2022 한국철도학회논문집 Vol.25 No.2

Recently, as operation patterns such as express and direct operation of subways in the metropolitan area become more and more complicated, a more systematic and optimized approach is needed to develop train schedules. Among them, a microscopic train operation simulation that imitates real railroad operation is being used in various fields such as feasibility analysis of train operation schedule, TPS analysis, and capacity evaluation in the construction planning and operation stages. In this study, going further than evaluating the feasibility of train schedules, we used a simulation to evaluate alternatives to a repetition plan by artificially generating delays in some trains and quantitatively calculating secondary delays as they spread through the train schedule, targeting common sections of the AnsanGwacheon Line and SuinBundang Line.