| The accurate probability estimation of passengers’ path selection, under railtransit barrier-free transfer mode, can help lines and stations withstand the realpressure of passenger traffic. It is of great significance for the projection of lines andstations, resources allocation and decision-making in the rail transit network. In themeantime, passing-by passenger flow is an important part of line passengers. It is thebasis of various of rail transit passenger analysis to assign the barrier-free transferpassengers to corresponding lines and stations. Currently, most of the relatedresearches focus on defining the optional multi-path weights, which will be combinedby several impact factors. Then the passenger route choice probability will bedetermined by the route weights. In this way, large number of passenger travel dataaccumulated in AFC (Automatic Fare Collection) system is not used and wasted whilethe mass transport passenger data contains valuable knowledge. This paper makes useof the passenger data from AFC system, try to find passenger traffic laws using datamining techniques, establish a new distribution model of rail transit barrier-freepassenger transfer mode. Also some key issues related to distribution underbarrier-free passenger transfer were studied.In this paper, the characteristics of rail passenger traffic were analyzed firstly;passenger flow in peak time interval is used as basis of passenger analysis accordingto the characteristics of rail transport. Standard travel time of OD (Origin, Destination)route is the reference for classification. The distribution mode is based on the ODcluster analysis of the actual passage of time, supplemented by introducing AttractWeights Coefficient method into passenger flow research in rail transit, and ultimatelythe whole routes of the barrier-free passenger transfer assigned. The distribution mode divides the optional routes of OD using K short paths algorithm and line routesalgorithm, splits optional route into atoms paths, combines the atoms paths standardtravel time and transfer time, then results into the OD standard travel time. For thoseroutes not available for cluster analysis, those paths that cannot be distinguished bystandard travel time, Attract Weights Coefficient method is introduced into passengerflow research in rail transit no-barrier transfer mode based on the bus OD matrixestimation model. The method uses passenger entry and exit data from AFC(automatic fare collection) system as input and passengers transfer times andpassing-by stations (distance factor) are also considered. Finally, a new passengerflow analysis model is presented. At last the distribution mode was verified based onthe passenger data from a running rail transit network. The case indicates that it canincrease the accuracy to consider the range of passenger flow peak. |
