The Urban Rail Transit Passengers’ Time-Space Extended Travel Path Matching Based On AFC Data And Train Timetable

With the development of urban rail transit networked operation, the choice of people’s travel is more diverse which causes passenger flow to present a more complex distribution. AFC system, that has a low cost of data acquisition, provides OD records of each passenger’s time-space information. AFC system and corresponding data mining technology make it possible to get passengers’OD data on the individual level from AFC system. In this context, based on rain schedules and AFC data and data mining, this study aims to quantize matching degree of all possible expansion of space-time travel route and the actual travel path to reproduce as closely as possible per passenger travel process which lay a foundation for passenger flow distribution simulation on the individual level.First of all, this paper describes the AFC data structure, the possible problems of original data and corresponding pretreatment methods. Through the analysis for each time component volatility in travel time, this paper divide travel time to pit time, vehicle time, transfer time and exit time which provides a basis for subsequent research.Secondly, this paper proposes time-space extended travel path search algorithm based on effective physical path. On the basis of physical topology network, this paper establishes time-space extended travel network which bases on AFC data and train timetable. On the basis of time-space extended travel network and effective physical path, this paper proposes a search algorithm which gives priority to exit arc constraint and judge pit arc constraint later for time-space extended travel path without transfer station. For time-space extended travel path with transfer station, this paper proposes a traversed search algorithm combined with the depth-first search algorithm which gives priority to constraint of pit arc and exit arc and judge transfer arc constraint later.After that, this paper puts forward a parameter, the matching probability, to quantize matching degree of all possible time-space extended travel path and the actual travel path, discusses its meaning and calculation method and design specific calculation process. It apply Logit random path selection probability model to verify the feasibility of calculation method of the matching probability. Based on matching probability, this paper use roulette method to match AFC data record on time-space extended travel path to so as to realize the simulation of passenger flow distribution, and establish modularization design of simulation system.Finally, this paper utilizes the AFC data to analyze time-temporal distribution characteristics of passenger flow of Line 5 of Beijing Metro comparatively. It take Tiantongyuan North station to Dongsishitiao station as typical commuter OD route for the study to estimate matching probability of passengers’time-space extended travel path in the morning peak hour 7:30-8:30, and calculate selection probability for each physical paths for all the passengers within the hour. Compared with value of logit models, the difference is small which proves the feasibility and effectiveness of matching probability calculation method proposed in this paper.