Research On The Problems Of Real-time Passenger Flow Measurement In Urban Transit Based On AFC System

The Urban railway construction of our country will meet its peak time during the"13th Five-Year" plan period.To fully achieve the advantage of the transport capacity and improve its service quality, the request of the urban railway operators and participants is getting higher and higher, which means there will be a great challenge to the collection and measurement of the passenger traffic information.However, the traditional monitoring and metro clearing systems have already could not meet the actual demands because of the relative lagging, while the video recognition techniques developed in recent years is not matched either as a result of its expensive equipment、low accuracy and recognition speed,which leads to the requirement of a new technology to measure the passenger flow real-time and prospectively. Therefore, the big data technology and traditional forecast method of the passenger flow were united together in this paper to research the key problems of real-time passenger flow measurement system in urban transit based on the AFC system. The system can accomplish the accurate measurement of passenger information by predicting the destinations of passengers and the passenger flow distribution once they enter the metro station with their smart cards, to support the optimization of the layout of station facilities 、detailed transportation organization and the selection of travelling routes.The main research contents of this paper are as follows:(1) Introduced the current monitoring system of urban transit passenger flow、presented the relevant studies of the road network topology、 method of solving paths and AFC data mining, lead to the research objectives and research contents.And after the analysis of the traditional passenger flow monitoring system, the four key problems of real-time passenger flow measurement system in urban transit based on AFC system were put forward.(2) Define a suitable effective path set based on the research demand, and abstract the urban transit network, establish the model of urban railway reasonable path set, realize the generating algorithm of effective path set by computer programming and use an example to check the algorithm.(3) The relative principles of ETL was used to clean the AFC data, after which the mining algorithm of the passenger destination was designed, finally developed a data mining engine software.Take AFC data of Chengdu metro as an example, which proved the accuracy and real-time of the engine, modified the problems appeared in the test, and then the generation of a short time passenger OD matrix was accomplished.(4) Verified that the travel time of passengers on a single path OD obeys lognormal distribution in view of the AFC historical data, then derived the distribution law of the travel time when there are multi paths, proposed the parameter estimation method, established the estimation model of the passenger distribution proportion among different paths. And verified the derived process of the theory based on the real data from AFC, finally we got the determining method of the passenger flow distribution when there is an obvious difference among the travelling time of different routes.While maximum utility theory was applied when the travelling times are similar.(5) Build the simulation development environment at the local-host to accomplish the measurement system of the urban railway passenger flow based on the AFC system, which achieved the real time measurement to the passenger information and provide its external interface to make the information spread conveniently.Load the AFC history data of Chengdu metro to test the software system, which turned out to be smooth and strong real-time of the passenger information, this proved the effectiveness and reliability of the models and algorithms in this paper.