Bus Passenger OD Matrix In-Ference Based On Multi-day IC Card Data And AVL Data

OD,the trip origin and destination of the bus passenger is important for the transit plan-ning,transit management and operation. While the traditional trip survey need much time and money,and only a small proportion of passengers can be involved. With the development of intelligent transportation system, the IC card data from automated fare-collection system(AFC) and the GPS data from automatic vehicle location system(AVL) record detailed trav-el information of bus passengers. Now instead of costly and infrequent traditional trip sur-vey, Origin-Destination (OD) matrix can be obtained from the AFC and AVL data by inte-grating them. In this paper, a new method is presented for OD matrix inference, and this method including four steps, namely origin inference,destination inference,interchange in-ference and final OD inference. To infer origin, this paper select the bus arrival time be-tween two stations as the time constraint,considering the bus is often crowed during peak hour in China and some improvement were also made on the origin inference of first and last station per run. Finally the method is applied to the IC card data and GPS data from Chengdu transit system and origins for 97.5% of IC card data were inferred.In destination inference, multi-day IC card data of passengers were used to improve the inference ratio of the passenger’s last trip stage and the single trip stage in one day. About 73.45% of transactions, destinations were inferred using smartcard transaction data of cur-rent day. After multi-day transaction data were used, destinations for another 2.69% of all transactions were inferred and finally about 76.4% of transactions, destinations were in-ferred. Besides, the result showed that a higher percentage of transactions’ destinations could be inferred on weekdays compared with weekends, and seniors own a higher per-centage of transactions with destination inferred compared with adults and students.The interchange inference based on the maximum interchange distance,the maximum interchange time and the maximum waiting time. The result of interchaneg inference shows that bus transfer coefficient is about 1.25, but to some degree compared with the week-ends,weekdays owns a higher bus transfer coefficient.Finally,we obtained the OD matrix by Oracle database and made an analysis on the spatial distribution of origin,destination and interchange station.